fix: adding date utils

* a simp0le scaffold

* feature: simple npm setup

* feature: testing setup and dummy scenarios

* chore: dumping kafak just via backend

* chore: dcleaning gitignore

* features: boilerplate fixtures and stuff

* test: extra tests

* chore: update the test suite to be callable via makefile

* chore: cleaning

* chore: updating interactions setup

* small cleaning

* chore: cleaning shitty code

.gitignore

@@ -11,3 +11,6 @@ paper/src/bib/auto
 experiments/airflow/logs/*
 experiments/airflow/logs/scheduler/
 experiments/airflow/logs/dag_processor_manager/
+tests/e2e/node_modules/**
+**/auto/*.el
+*.old

Makefile

@@ -11,46 +11,74 @@ PYTEST    := $(VENV)/bin/pytest
 
 .DEFAULT_GOAL := help
 
-all: pdf
+.PHONY: help
-
+help:
-run.webapp:
+	@echo "pdf.build pdf.watch pdf.clean | test.backend test.e2e test.all | web.dev | install | stats.lines"
-	@cd web && npm install && npm run dev
 
 $(BUILDDIR):
 	mkdir -p paper/$(BUILDDIR)
 
-pdf: $(BUILDDIR)
+.PHONY: pdf.build
-	@echo "Concatenating source code..."
+pdf.build: $(BUILDDIR)
 	@bash paper/concat_code.sh
 	@cd $(SRCDIR) && \
 	$(LATEXMK) -pdf -jobname=$(JOBNAME) \
 		-interaction=nonstopmode -file-line-error \
 		-outdir=../$(BUILDDIR) $(TEX)
 
-watch: $(BUILDDIR)
+.PHONY: pdf.watch
+pdf.watch: $(BUILDDIR)
 	@cd $(SRCDIR) && \
 	$(LATEXMK) -pvc -pdf -jobname=$(JOBNAME) \
 		-interaction=nonstopmode -file-line-error \
 		-r ../.latexmkrc \
 		-outdir=../$(BUILDDIR) $(TEX)
 
-clean:
+.PHONY: pdf.clean
+pdf.clean:
 	@cd $(SRCDIR) && \
 	$(LATEXMK) -C -jobname=$(JOBNAME) -outdir=../$(BUILDDIR) || true
 	rm -rf paper/$(BUILDDIR)/*
 
+.PHONY: test.backend
+test.backend: $(VENV)
+	$(PYTEST) -v
+
+.PHONY: test.e2e
+test.e2e:
+	@cd tests/e2e && npm install
+	@cd tests/e2e && npx playwright install chromium
+	@test -f tests/e2e/.env || cp tests/e2e/.env.example tests/e2e/.env
+	@timeout 30 bash -c 'until curl -sf http://localhost:5000/health > /dev/null 2>&1; do sleep 1; done' || (echo "Backend not ready" && exit 1)
+	@timeout 30 bash -c 'until curl -sf http://localhost:3000 > /dev/null 2>&1; do sleep 1; done' || (echo "Web app not ready" && exit 1)
+	@timeout 30 bash -c 'until curl -sf http://localhost:8085/health > /dev/null 2>&1; do sleep 1; done' || (echo "Airflow not ready" && exit 1)
+	@cd tests/e2e && npm test
+
+.PHONY: test.all
+test.all: test.backend test.e2e
+
+.PHONY: web.dev
+web.dev:
+	@cd web && npm install && npm run dev
+
 $(VENV):
 	python3 -m venv $(VENV)
 	$(PIP) install --upgrade pip
 
+.PHONY: install
 install: $(VENV)
 	$(PIP) install -r requirements.txt
 
-test: $(VENV)
+.PHONY: stats.lines
-	$(PYTEST) -v
+stats.lines:
-
-count-lines:
 	@find . \( -path '*/node_modules' -o -path '*/.venv' -o -path '*/venv' \) -prune -o \
 	\( -name "*.ts" -o -name "*.py" \) -type f -print0 | xargs -0 cat | wc -l
 
-.PHONY: all pdf clean watch run.webapp install test
+.PHONY: pdf clean watch run.webapp test count-lines all
+pdf: pdf.build
+clean: pdf.clean
+watch: pdf.watch
+run.webapp: web.dev
+test: test.backend
+count-lines: stats.lines
+all: pdf.build

README.md

@@ -1,5 +1,12 @@
+<img width="200" align="left" src="https://github.com/user-attachments/assets/d148b00d-e9f9-4280-89cc-0cc866e17251" />
+
+### PHANTOM
+
 [![Build PDF](https://github.com/velocitatem/PHANTOM/actions/workflows/latex.yml/badge.svg)](https://github.com/velocitatem/PHANTOM/actions/workflows/latex.yml)
+[![TPU Research Cloud](https://img.shields.io/badge/TPU%20Research%20Cloud-TRC%20supported-4285F4?logo=googlecloud&logoColor=white)](https://sites.research.google/trc/faq/)
+[![Vercel Deploy](https://deploy-badge.vercel.app/?url=https://phantom-hotel.vercel.app&name=Hotel)](https://phantom-hotel.vercel.app)
+[![Vercel Deploy](https://deploy-badge.vercel.app/?url=https://phantom-airline.vercel.app&name=Airline)](https://phantom-airline.vercel.app)
+
+
 
-- https://phantom-hotel.vercel.app/
-- https://phantom-airline.vercel.app/
 

backend/provider/app.py

@@ -19,11 +19,11 @@ from procesing.pricers import (
     ElasticityBasedPricer
 )
 from procesing.steps import (
-    StateSpace,
     PredictPricesStep
 )
 from procesing import PipelineContext
 sys.path.append(os.path.dirname(os.path.abspath(__file__))+ "/../../lib/")
+print(os.path.dirname(os.path.abspath(__file__))+ "/../../lib/")
 from lib.model_registry import ModelRegistry
 
 # Config
@@ -47,122 +47,52 @@ def health() -> dict:
 
 @app.get("/api/{mode}/price/{productId}", response_model=PriceResponse)
 def get_price(mode: Literal['hotel', 'airline'], productId: str, sessionId: Optional[str] = Query(None), experimentId: Optional[str] = Query(None)):
+    """
+    THIS is the fast lookup service (mechanism).
+    Priority: session-keyed price > global optimal price > base price
+    """
     product = supabase.table(f'{mode}_products').select("metadata").eq('id', productId).execute().data[0]
     if not product: raise HTTPException(404, f"Product {productId} not found")
 
     metadata = product['metadata']
     base_price = metadata.get('base_price', 100.0)
 
-    class Provider(SupabaseProvider, BackendAPIProvider):
+    # PRIORITY 1: session-aware price (computed by Airflow worker)
-        def __init__(self, backend_url: str):
+    if sessionId:
-            SupabaseProvider.__init__(self)
+        session_price = registry.get_session_price(sessionId, productId)
-            BackendAPIProvider.__init__(self, backend_url=backend_url)
+        if session_price is not None:
-
-    context = PipelineContext(
-        provider=Provider(backend_url=os.getenv("BACKEND_URL")),
-        store_mode=mode
-    )
-
-    pricing_model = registry.get_pricing_model('latest')
-    elasticity_df = registry.get_elasticity('latest')
-
-    if pricing_model is None or elasticity_df is None:
             return PriceResponse(
                 productId=productId,
-            price=base_price,
+                price=session_price,
                 base_price=base_price,
-            markup=1.0,
+                markup=session_price/base_price,
-            elasticity=None
+                elasticity=None,
+                model_version='session-aware'
             )
 
-    products = context.products
+    # PRIORITY 2: global pre-computed prices (surge pricing)
-    if products.empty:
+    prices_df = registry.get_prices('latest')
-        raise HTTPException(500, "No products available in catalog")
+    if prices_df is not None:
-
-    # merge elasticity with product base prices
-    products_with_meta = products.copy()
-    products_with_meta['base_price'] = products_with_meta['metadata'].apply(
-        lambda m: m.get('base_price', 100.0) if isinstance(m, dict) else 100.0
-    )
-
-    merged = products_with_meta[['id', 'base_price']].rename(
-        columns={'id': 'productId'}
-    ).merge(
-        elasticity_df[['productId', 'elasticity']],
-        on='productId',
-        how='left'
-    ).fillna({'elasticity': 0.0})
-
-    # compute demand: use pricer's mean_demand if available, else default
-    demand_values = (pricing_model.mean_demand
-                    if hasattr(pricing_model, 'mean_demand') and pricing_model.mean_demand is not None
-                    else np.ones(len(merged)) * 10.0)
-
-    # build state space with session features if sessionId provided
-    session_features = pd.DataFrame()
-    if sessionId:
-        try:
-            # fetch recent session interactions from backend
-            from procesing.steps.session import ExtractSessionFeaturesStep
-            import requests
-            from datetime import datetime, timedelta
-
-            t_end = datetime.utcnow()
-            t_start = t_end - timedelta(hours=1)
-            backend_url = os.getenv("BACKEND_URL")
-            print(backend_url)
-
-            resp = requests.get(
-                f"{os.getenv('BACKEND_URL')}/api/kafka/dump", # TODO: THIS IS SHIT, must fix this
-                params={'topic': 'user-interactions', 't_start': t_start.isoformat(), 't_end': t_end.isoformat()},
-                timeout=2
-            )
-
-            if resp.ok:
-                msgs = resp.json().get('messages', [])
-                interactions_df = pd.DataFrame(msgs)
-
-                if not interactions_df.empty and 'sessionId' in interactions_df.columns:
-                    session_interactions = interactions_df[interactions_df['sessionId'] == sessionId]
-
-                    if not session_interactions.empty:
-                        extractor = ExtractSessionFeaturesStep(context=context)
-                        session_features_df = extractor.transform(session_interactions)
-
-                        if not session_features_df.empty:
-                            session_features = session_features_df.drop(columns=['sessionId'])
-        except Exception as e:
-            print(f"[session-features-error] {e}")
-            # continue without session features
-
-    state = StateSpace(
-        demand=demand_values,
-        prices=merged['base_price'].values,
-        session_features=session_features,
-        product_ids=merged['productId'].values,
-        elasticity=merged['elasticity'].values,
-        metadata={'sessionId': sessionId, 'experimentId': experimentId}
-    )
-
-    oracle = PredictPricesStep(context=context)
-    prices_df = oracle.transform((pricing_model, state))
-
         product_price_row = prices_df[prices_df['productId'] == productId]
-    if product_price_row.empty:
+        if not product_price_row.empty:
-        raise HTTPException(404, f"No pricing available for product {productId}")
+            optimal_price = float(product_price_row['optimal_price'].iloc[0])
-
-    optimal_price = float(product_price_row['predicted_price'].iloc[0])
-
-    product_elasticity_row = elasticity_df[elasticity_df['productId'] == productId]
-    product_elasticity = (float(product_elasticity_row['elasticity'].iloc[0])
-                         if not product_elasticity_row.empty else None)
-
             return PriceResponse(
                 productId=productId,
                 price=optimal_price,
                 base_price=base_price,
                 markup=optimal_price/base_price,
-        elasticity=product_elasticity
+                elasticity=None,
+                model_version='surge'
+            )
+
+    # PRIORITY 3: fallback to base price
+    return PriceResponse(
+        productId=productId,
+        price=base_price,
+        base_price=base_price,
+        markup=1.0,
+        elasticity=None,
+        model_version='base'
     )
 
 @app.get("/models")

backend/provider/requirements.txt

@@ -12,4 +12,5 @@ graphviz
 python-dotenv>=1.0.0
 requests>=2.31.0
 typing-extensions>=4.8.0
-pickle5>=0.0.11; python_version < '3.8'
+pypickle
+pymc

backend/server/app.py

@@ -198,12 +198,16 @@ def dump_logs(
             auto_offset_reset='earliest',
             enable_auto_commit=False,
             value_deserializer=lambda x: json.loads(x.decode('utf-8')),
-            consumer_timeout_ms=5000
+            consumer_timeout_ms=30000,
+            fetch_max_wait_ms=10000,
+            max_poll_records=1000
         )
 
         events = []
         for msg in consumer:
             events.append(msg.value)
+            if last_n and len(events) >= last_n * 2:
+                break
 
         consumer.close()
 
@@ -290,6 +294,7 @@ async def get_products(
         query = supabase.table(table).select('*')
 
         # filter by exact date_index if provided
+        # dateIndex from frontend is days from today, convert to days since epoch
         if dateIndex is not None:
             query = query.eq('date_index', dateIndex)
 

docker-compose.yml

@@ -1,4 +1,24 @@
 services:
+  tensorboard-rl:
+    image: tensorflow/tensorflow:latest
+    container_name: "PHANTOM-tensorboard-rl"
+    ports:
+      - "6007:6006"
+    volumes:
+      - ./sim/rl/runs:/logs
+    command: tensorboard --logdir=/logs --host=0.0.0.0 --port=6006
+    restart: unless-stopped
+
+  tensorboard-ml:
+    image: tensorflow/tensorflow:latest
+    container_name: "PHANTOM-tensorboard-ml"
+    ports:
+      - "6006:6006"
+    volumes:
+      - ./experiments/ml/runs:/logs
+    command: tensorboard --logdir=/logs --host=0.0.0.0 --port=6006
+    restart: unless-stopped
+
   backend:
     container_name: "PHANTOM-backend"
     build:
@@ -92,23 +112,20 @@ services:
     depends_on:
       - postgres
     environment:
-      - AIRFLOW__CORE__EXECUTOR=SequentialExecutor
+      - AIRFLOW__CORE__EXECUTOR=LocalExecutor
       - AIRFLOW__DATABASE__SQL_ALCHEMY_CONN=postgresql+psycopg2://airflow:airflow@postgres/airflow
       - AIRFLOW__CORE__FERNET_KEY=${AIRFLOW_FERNET_KEY}
       - AIRFLOW__CORE__LOAD_EXAMPLES=false
       - AIRFLOW__CORE__ENABLE_XCOM_PICKLING=true
+      - AIRFLOW__CORE__PARALLELISM=16
+      - AIRFLOW__CORE__DAG_CONCURRENCY=8
+      - AIRFLOW__CORE__MAX_ACTIVE_RUNS_PER_DAG=4
       - _AIRFLOW_DB_MIGRATE=true
       - _AIRFLOW_WWW_USER_CREATE=true
       - _AIRFLOW_WWW_USER_USERNAME=admin
       - _AIRFLOW_WWW_USER_PASSWORD=admin
       - REDIS_HOST=redis
       - REDIS_PORT=6379
-    volumes:
-      - ./experiments/airflow/dags:/opt/airflow/dags
-      - ./experiments/airflow/logs:/opt/airflow/logs
-      - ./experiments/airflow/plugins:/opt/airflow/plugins
-      - ./experiments/procesing:/opt/airflow/procesing
-      - ./lib:/opt/airflow/lib
     command: version
     restart: "no"
 
@@ -122,13 +139,20 @@ services:
       - airflow-init
       - redis
     environment:
-      - AIRFLOW__CORE__EXECUTOR=SequentialExecutor
+      - AIRFLOW__CORE__EXECUTOR=LocalExecutor
       - AIRFLOW__DATABASE__SQL_ALCHEMY_CONN=postgresql+psycopg2://airflow:airflow@postgres/airflow
       - AIRFLOW__CORE__FERNET_KEY=${AIRFLOW_FERNET_KEY}
       - AIRFLOW__CORE__DAGS_ARE_PAUSED_AT_CREATION=true
       - AIRFLOW__CORE__LOAD_EXAMPLES=false
       - AIRFLOW__CORE__ENABLE_XCOM_PICKLING=true
+      - AIRFLOW__CORE__PARALLELISM=16
+      - AIRFLOW__CORE__DAG_CONCURRENCY=8
+      - AIRFLOW__CORE__MAX_ACTIVE_RUNS_PER_DAG=4
+      - AIRFLOW__SCHEDULER__MIN_FILE_PROCESS_INTERVAL=30
+      - AIRFLOW__SCHEDULER__DAG_DIR_LIST_INTERVAL=60
       - AIRFLOW__WEBSERVER__EXPOSE_CONFIG=true
+      - AIRFLOW__WEBSERVER__SECRET_KEY=${AIRFLOW_SECRET_KEY}
+      - AIRFLOW__API__AUTH_BACKENDS=airflow.api.auth.backend.basic_auth
       - KAFKA_HOST=kafka
       - KAFKA_PORT=29092
       - BACKEND_URL=http://backend:5000
@@ -138,12 +162,6 @@ services:
       - REDIS_PORT=6379
     ports:
       - "${AIRFLOW_WEBSERVER_PORT:-8085}:8080"
-    volumes:
-      - ./experiments/airflow/dags:/opt/airflow/dags:ro
-      - ./experiments/airflow/logs:/opt/airflow/logs
-      - ./experiments/airflow/plugins:/opt/airflow/plugins:ro
-      - ./experiments/procesing:/opt/airflow/procesing:ro
-      - ./lib:/opt/airflow/lib:ro
     command: webserver
     restart: unless-stopped
     healthcheck:
@@ -164,12 +182,20 @@ services:
       redis:
         condition: service_started
     environment:
-      - AIRFLOW__CORE__EXECUTOR=SequentialExecutor
+      - AIRFLOW__CORE__EXECUTOR=LocalExecutor
       - AIRFLOW__DATABASE__SQL_ALCHEMY_CONN=postgresql+psycopg2://airflow:airflow@postgres/airflow
       - AIRFLOW__CORE__FERNET_KEY=${AIRFLOW_FERNET_KEY}
       - AIRFLOW__CORE__DAGS_ARE_PAUSED_AT_CREATION=true
       - AIRFLOW__CORE__LOAD_EXAMPLES=false
       - AIRFLOW__CORE__ENABLE_XCOM_PICKLING=true
+      - AIRFLOW__CORE__PARALLELISM=16
+      - AIRFLOW__CORE__DAG_CONCURRENCY=8
+      - AIRFLOW__CORE__MAX_ACTIVE_RUNS_PER_DAG=4
+      - AIRFLOW__SCHEDULER__MIN_FILE_PROCESS_INTERVAL=30
+      - AIRFLOW__SCHEDULER__DAG_DIR_LIST_INTERVAL=60
+      - AIRFLOW__SCHEDULER__PARSING_PROCESSES=2
+      - AIRFLOW__WEBSERVER__SECRET_KEY=${AIRFLOW_SECRET_KEY}
+      - AIRFLOW__API__AUTH_BACKENDS=airflow.api.auth.backend.basic_auth
       - KAFKA_HOST=kafka
       - KAFKA_PORT=29092
       - BACKEND_URL=http://backend:5000
@@ -177,12 +203,6 @@ services:
       - NEXT_PUBLIC_SUPABASE_ANON_KEY=${NEXT_PUBLIC_SUPABASE_ANON_KEY}
       - REDIS_HOST=redis
       - REDIS_PORT=6379
-    volumes:
-      - ./experiments/airflow/dags:/opt/airflow/dags:ro
-      - ./experiments/airflow/logs:/opt/airflow/logs
-      - ./experiments/airflow/plugins:/opt/airflow/plugins:ro
-      - ./experiments/procesing:/opt/airflow/procesing:ro
-      - ./lib:/opt/airflow/lib:ro
     command: scheduler
     restart: unless-stopped
     healthcheck:
@@ -208,13 +228,9 @@ services:
       - KAFKA_PORT=29092
       - NEXT_PUBLIC_SUPABASE_URL=${NEXT_PUBLIC_SUPABASE_URL}
       - NEXT_PUBLIC_SUPABASE_ANON_KEY=${NEXT_PUBLIC_SUPABASE_ANON_KEY}
+      - BACKEND_URL=http://localhost:5000
     ports:
       - "${PROVIDER_PORT:-5001}:5001"
-    volumes:
-      - ./lib:/app/lib:ro
-      - ./experiments/procesing:/app/procesing:ro
-      - ./backend/provider:/app/provider:ro
-    command: python -m uvicorn provider.app:app --host 0.0.0.0 --port 5001
     restart: unless-stopped
 
 volumes:

docker/Airflow.dockerfile

@@ -21,3 +21,10 @@ RUN pip install --no-cache-dir \
 
 # set airflow home
 ENV AIRFLOW_HOME=/opt/airflow
+
+COPY --chown=airflow:root experiments/airflow/dags ${AIRFLOW_HOME}/dags
+COPY --chown=airflow:root experiments/procesing ${AIRFLOW_HOME}/procesing
+COPY --chown=airflow:root lib ${AIRFLOW_HOME}/lib
+
+# create logs and plugins dirs (airflow expects them)
+RUN mkdir -p ${AIRFLOW_HOME}/logs ${AIRFLOW_HOME}/plugins

docker/Airflow.railway.dockerfile

@@ -0,0 +1,41 @@
+FROM apache/airflow:2.7.3-python3.11
+
+USER root
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    supervisor \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+USER airflow
+
+COPY requirements.txt /tmp/requirements.txt
+RUN pip install --no-cache-dir -r /tmp/requirements.txt
+
+RUN pip install --no-cache-dir \
+    psycopg2-binary \
+    apache-airflow-providers-postgres
+
+ENV AIRFLOW_HOME=/opt/airflow
+ENV AIRFLOW__CORE__EXECUTOR=SequentialExecutor
+ENV AIRFLOW__CORE__LOAD_EXAMPLES=false
+ENV AIRFLOW__CORE__ENABLE_XCOM_PICKLING=true
+ENV AIRFLOW__WEBSERVER__EXPOSE_CONFIG=true
+
+# copy all code into image (standalone - no volume mounts needed)
+COPY --chown=airflow:root experiments/airflow/dags ${AIRFLOW_HOME}/dags
+COPY --chown=airflow:root experiments/procesing ${AIRFLOW_HOME}/procesing
+COPY --chown=airflow:root lib ${AIRFLOW_HOME}/lib
+
+RUN mkdir -p ${AIRFLOW_HOME}/logs ${AIRFLOW_HOME}/plugins
+
+# copy entrypoint script
+COPY --chown=airflow:root docker/airflow-railway-entrypoint.sh /entrypoint.sh
+USER root
+RUN chmod +x /entrypoint.sh
+USER airflow
+
+EXPOSE 8080
+
+ENTRYPOINT ["/entrypoint.sh"]

docker/Provider.dockerfile

@@ -14,11 +14,13 @@ RUN apt-get update && apt-get install -y \
 COPY backend/provider/requirements.txt /app/
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Structure will be mounted via volumes:
+# Copy application code into image
-# /app/lib -> lib/
+COPY lib/ /app/lib/
-# /app/procesing -> experiments/procesing/
+COPY experiments/procesing/ /app/procesing/
-# /app/provider -> backend/provider/
+COPY backend/provider/ /app/provider/
 
 ENV PYTHONPATH=/app:/app/lib:/app/procesing
 
-CMD ["python", "-m", "uvicorn", "provider.app:app", "--host", "0.0.0.0", "--port", "5001"]
+WORKDIR /app/provider
+
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "5001"]

docker/airflow-railway-entrypoint.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+set -e
+
+# init db and create admin user on first run
+airflow db migrate
+
+# create admin user if not exists
+airflow users create \
+    --username "${AIRFLOW_ADMIN_USER:-admin}" \
+    --password "${AIRFLOW_ADMIN_PASSWORD:-admin}" \
+    --firstname Admin \
+    --lastname User \
+    --role Admin \
+    --email admin@example.com || true
+
+# start scheduler in background
+airflow scheduler &
+
+# start webserver in foreground (Railway needs one foreground process)
+exec airflow webserver --port ${PORT:-8080}

experiments/airflow/dags/elasticity_pricing_dag.py

@@ -1,346 +0,0 @@
-from airflow import DAG
-from airflow.operators.python import PythonOperator
-from airflow.utils.dates import days_ago
-from datetime import timedelta
-import pandas as pd
-import logging
-import sys
-import pickle
-import io
-
-# add parent dir to path so procesing package can be imported
-sys.path.insert(0, '/opt/airflow')
-
-from procesing.context import PipelineContext
-from procesing.providers import SupabaseProvider, BackendAPIProvider
-from procesing.steps import (
-    FetchInteractionsStep,
-    FetchPriceLogsStep,
-    CreatePriceBucketsStep,
-    AugmentEventNamesStep,
-    ChunkByTimeWindowStep,
-    ComputeDemandForChunksStep,
-    AggregatePriceLogsStep,
-    ComputeElasticityStep,
-    BuildStateSpaceStep,
-    FitPricingFunctionStep,
-    PredictPricesStep,
-)
-
-default_args = {
-    'owner': 'phantom-research',
-    'depends_on_past': False,
-    'email_on_failure': False,
-    'email_on_retry': False,
-    'retries': 2,
-    'retry_delay': timedelta(minutes=5),
-}
-
-def get_provider():
-    """Factory to create composite provider"""
-    class CompositeProvider(SupabaseProvider, BackendAPIProvider):
-        def __init__(self):
-            SupabaseProvider.__init__(self)
-            BackendAPIProvider.__init__(self)
-    return CompositeProvider()
-
-def get_context(**kwargs):
-    """Build pipeline context from Airflow config"""
-    dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
-    return PipelineContext(
-        provider=get_provider(),
-        store_mode=dag_conf.get('store_mode', 'hotel'),
-        window_size=dag_conf.get('window_size', '30s'),
-        n_price_buckets=dag_conf.get('n_price_buckets', 5),
-        elasticity_method=dag_conf.get('elasticity_method', 'point'),
-        min_observations=dag_conf.get('min_observations', 2),
-    )
-
-# atomic task functions (each wraps one sklearn step)
-def fetch_interactions(**kwargs):
-    """Task: Fetch interaction data from Kafka"""
-    context = get_context(**kwargs)
-    step = FetchInteractionsStep(context)
-    df = step.transform(None)
-
-    kwargs['ti'].xcom_push(key='interactions_raw', value=pickle.dumps(df))
-    logging.info(f"Fetched {len(df)} interaction records")
-    return len(df)
-
-def fetch_price_logs(**kwargs):
-    """Task: Fetch price logs from Kafka"""
-    context = get_context(**kwargs)
-    step = FetchPriceLogsStep(context)
-    df = step.transform(None)
-
-    kwargs['ti'].xcom_push(key='price_logs_raw', value=pickle.dumps(df))
-    logging.info(f"Fetched {len(df)} price records")
-    return len(df)
-
-def create_price_buckets(**kwargs):
-    """Task: Create price buckets for interactions"""
-    ti = kwargs['ti']
-    df = pickle.loads(ti.xcom_pull(key='interactions_raw'))
-
-    context = get_context(**kwargs)
-    step = CreatePriceBucketsStep(context)
-    df = step.transform(df)
-
-    ti.xcom_push(key='interactions_bucketed', value=pickle.dumps(df))
-    logging.info(f"Created price buckets for {len(df)} interactions")
-    return len(df)
-
-def augment_event_names(**kwargs):
-    """Task: Augment event names with product and price schema"""
-    ti = kwargs['ti']
-    df = pickle.loads(ti.xcom_pull(key='interactions_bucketed'))
-
-    context = get_context(**kwargs)
-    step = AugmentEventNamesStep(context)
-    df = step.transform(df)
-
-    ti.xcom_push(key='interactions_final', value=pickle.dumps(df))
-    logging.info(f"Augmented event names for {len(df)} interactions")
-    return len(df)
-
-def chunk_interactions(**kwargs):
-    """Task: Chunk interactions into time windows"""
-    ti = kwargs['ti']
-    df = pickle.loads(ti.xcom_pull(key='interactions_final'))
-
-    context = get_context(**kwargs)
-    step = ChunkByTimeWindowStep(context)
-    chunks = step.transform(df)
-
-    ti.xcom_push(key='interaction_chunks', value=pickle.dumps(chunks))
-    logging.info(f"Generated {len(chunks)} interaction chunks")
-    return len(chunks)
-
-def compute_demand(**kwargs):
-    """Task: Compute demand vectors for all chunks"""
-    ti = kwargs['ti']
-    chunks = pickle.loads(ti.xcom_pull(key='interaction_chunks'))
-
-    context = get_context(**kwargs)
-    step = ComputeDemandForChunksStep(context)
-    demand_chunks = step.transform(chunks)
-
-    ti.xcom_push(key='demand_chunks', value=pickle.dumps(demand_chunks))
-    logging.info(f"Computed demand for {len(demand_chunks)} chunks")
-    return len(demand_chunks)
-
-def aggregate_price_logs(**kwargs):
-    """Task: Aggregate price logs into time windows """
-    ti = kwargs['ti']
-    df = pickle.loads(ti.xcom_pull(key='price_logs_raw'))
-
-    context = get_context(**kwargs)
-    step = AggregatePriceLogsStep(context)
-    price_chunks = step.transform(df)
-
-    ti.xcom_push(key='price_chunks', value=pickle.dumps(price_chunks))
-    logging.info(f"Aggregated {len(price_chunks)} price chunks")
-    return len(price_chunks)
-
-def compute_elasticity(**kwargs):
-    """Task: Compute price elasticity from demand and price chunks"""
-    ti = kwargs['ti']
-    demand_chunks = pickle.loads(ti.xcom_pull(key='demand_chunks'))
-    price_chunks = pickle.loads(ti.xcom_pull(key='price_chunks'))
-
-    context = get_context(**kwargs)
-    step = ComputeElasticityStep(context)
-    elasticity_df = step.transform((demand_chunks, price_chunks))
-
-    ti.xcom_push(key='elasticity_results', value=pickle.dumps(elasticity_df))
-    logging.info(f"Computed elasticity for {len(elasticity_df)} products")
-
-    return {
-        'n_products': len(elasticity_df),
-        'mean_elasticity': float(elasticity_df['elasticity'].mean()),
-        'median_elasticity': float(elasticity_df['elasticity'].median())
-    }
-
-def build_state_space(**kwargs):
-    """Task: Build state space from elasticity"""
-    ti = kwargs['ti']
-    elasticity_df = pickle.loads(ti.xcom_pull(key='elasticity_results'))
-
-    context = get_context(**kwargs)
-    step = BuildStateSpaceStep(context)
-    state_space = step.transform(elasticity_df)
-
-    ti.xcom_push(key='state_space', value=pickle.dumps(state_space))
-    logging.info("Built state space for pricing")
-    return True
-
-def fit_pricing_function(**kwargs):
-    """Task: Fit pricing function using elasticity"""
-    ti = kwargs['ti']
-    elasticity_df = pickle.loads(ti.xcom_pull(key='elasticity_results'))
-
-    context = get_context(**kwargs)
-    step = FitPricingFunctionStep(context)
-    pricer = step.transform(elasticity_df)
-
-    ti.xcom_push(key='pricer', value=pickle.dumps(pricer))
-    logging.info("Fitted pricing function")
-    return True
-
-def predict_prices(**kwargs):
-    """Task: Predict optimal prices"""
-    ti = kwargs['ti']
-    pricer = pickle.loads(ti.xcom_pull(key='pricer'))
-    state_space = pickle.loads(ti.xcom_pull(key='state_space'))
-
-    context = get_context(**kwargs)
-    step = PredictPricesStep(context)
-    prices_df = step.transform((pricer, state_space))
-
-    ti.xcom_push(key='predicted_prices', value=pickle.dumps(prices_df))
-    logging.info(f"Predicted prices for {len(prices_df)} products")
-    return len(prices_df)
-
-def publish_results(**kwargs):
-    """Task: Publish elasticity and pricing results to model registry"""
-    ti = kwargs['ti']
-    elasticity_df = pickle.loads(ti.xcom_pull(key='elasticity_results'))
-    prices_df = pickle.loads(ti.xcom_pull(key='predicted_prices'))
-
-    sys.path.insert(0, '/opt/airflow')
-    from lib.model_registry import ModelRegistry
-
-    registry = ModelRegistry()
-    dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
-
-    metadata = {
-        'timestamp': pd.Timestamp.now().isoformat(),
-        'window_size': dag_conf.get('window_size', '30s'),
-        'store_mode': dag_conf.get('store_mode', 'hotel'),
-        'dag_run_id': kwargs['dag_run'].run_id if kwargs.get('dag_run') else 'manual'
-    }
-
-    registry.publish_elasticity(elasticity_df, model_name='latest', metadata=metadata)
-
-    # get fitted pricer from XCom
-    pricer = pickle.loads(ti.xcom_pull(key='pricer'))
-    registry.publish_pricing_model(
-        pricer,
-        model_name='latest',
-        metadata={**metadata, 'model_type': type(pricer).__name__}
-    )
-
-    logging.info(f"Published elasticity + pricing for {len(elasticity_df)} products")
-
-    return {
-        'n_products': len(elasticity_df),
-        'registry_status': 'success',
-        'elasticity_mean': float(elasticity_df['elasticity'].mean())
-    }
-
-
-# DAG definition
-with DAG(
-    'elasticity_pricing_pipeline',
-    default_args=default_args,
-    description='E2E refactored pipeline: atomic steps with proper separation',
-    schedule_interval='*/15 * * * *',
-    start_date=days_ago(1),
-    catchup=False,
-    max_active_runs=1,
-    tags=['pricing', 'elasticity', 'research', 'refactored'],
-) as dag:
-
-    # parallel data fetching
-    t_fetch_interactions = PythonOperator(
-        task_id='fetch_interactions',
-        python_callable=fetch_interactions,
-        provide_context=True,
-    )
-
-    t_fetch_price_logs = PythonOperator(
-        task_id='fetch_price_logs',
-        python_callable=fetch_price_logs,
-        provide_context=True,
-    )
-
-    # interaction processing branch
-    t_create_buckets = PythonOperator(
-        task_id='create_price_buckets',
-        python_callable=create_price_buckets,
-        provide_context=True,
-    )
-
-    t_augment_events = PythonOperator(
-        task_id='augment_event_names',
-        python_callable=augment_event_names,
-        provide_context=True,
-    )
-
-    t_chunk_interactions = PythonOperator(
-        task_id='chunk_interactions',
-        python_callable=chunk_interactions,
-        provide_context=True,
-    )
-
-    t_compute_demand = PythonOperator(
-        task_id='compute_demand',
-        python_callable=compute_demand,
-        provide_context=True,
-    )
-
-    # price processing branch (VECTORIZED)
-    t_aggregate_prices = PythonOperator(
-        task_id='aggregate_price_logs',
-        python_callable=aggregate_price_logs,
-        provide_context=True,
-    )
-
-    # convergence: compute elasticity
-    t_compute_elasticity = PythonOperator(
-        task_id='compute_elasticity',
-        python_callable=compute_elasticity,
-        provide_context=True,
-    )
-
-    # pricing tasks
-    t_build_state = PythonOperator(
-        task_id='build_state_space',
-        python_callable=build_state_space,
-        provide_context=True,
-    )
-
-    t_fit_pricer = PythonOperator(
-        task_id='fit_pricing_function',
-        python_callable=fit_pricing_function,
-        provide_context=True,
-    )
-
-    t_predict_prices = PythonOperator(
-        task_id='predict_prices',
-        python_callable=predict_prices,
-        provide_context=True,
-    )
-
-    # publish to registry
-    t_publish = PythonOperator(
-        task_id='publish_results',
-        python_callable=publish_results,
-        provide_context=True,
-    )
-
-    # dependency graph (clear atomic flow)
-    # parallel fetches
-    [t_fetch_interactions, t_fetch_price_logs]
-
-    # interaction branch: fetch -> bucket -> augment -> chunk -> demand
-    t_fetch_interactions >> t_create_buckets >> t_augment_events >> t_chunk_interactions >> t_compute_demand
-
-    # price branch: fetch -> aggregate (vectorized)
-    t_fetch_price_logs >> t_aggregate_prices
-
-    # convergence: both branches -> elasticity
-    [t_compute_demand, t_aggregate_prices] >> t_compute_elasticity
-
-    # pricing: elasticity -> state + fit -> predict -> publish
-    t_compute_elasticity >> [t_build_state, t_fit_pricer] >> t_predict_prices >> t_publish

experiments/airflow/dags/ml_training_pipeline.py

@@ -0,0 +1,115 @@
+from airflow import DAG, Dataset
+from airflow.decorators import task
+from airflow.utils.dates import days_ago
+from datetime import timedelta
+import pandas as pd
+import logging
+import sys
+import pickle
+
+sys.path.insert(0, '/opt/airflow')
+
+from procesing.context import PipelineContext
+from procesing.providers import SupabaseProvider, BackendAPIProvider
+from procesing.steps import (
+    FetchInteractionsStep,
+    ValidateDataStep,
+    ExtractSessionFeaturesStep,
+    JoinLabelsStep,
+)
+
+TRAINING_DATASET = Dataset('phantom://ml/training-data')
+
+DEFAULT_ARGS = {
+    'owner': 'phantom-research',
+    'depends_on_past': False,
+    'email_on_failure': False,
+    'email_on_retry': False,
+    'retries': 2,
+    'retry_delay': timedelta(minutes=5),
+}
+
+
+class CompositeProvider(SupabaseProvider, BackendAPIProvider):
+    def __init__(self):
+        SupabaseProvider.__init__(self)
+        BackendAPIProvider.__init__(self)
+
+
+def _get_context(store_mode: str = 'hotel') -> PipelineContext:
+    return PipelineContext(provider=CompositeProvider(), store_mode=store_mode)
+
+
+with DAG(
+    'ml_training_pipeline',
+    default_args=DEFAULT_ARGS,
+    description='ML training data pipeline: fetch -> validate -> extract features -> label -> publish',
+    schedule=None,
+    start_date=days_ago(1),
+    catchup=False,
+    max_active_runs=1,
+    tags=['ml', 'training', 'features', 'research'],
+) as dag:
+
+    @task
+    def fetch_interactions(**kwargs) -> bytes:
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+        ctx = _get_context(dag_conf.get('store_mode', 'hotel'))
+        df = FetchInteractionsStep(ctx).transform(None)
+        logging.info(f"Fetched {len(df)} interactions, {df['sessionId'].nunique()} sessions")
+        return pickle.dumps(df)
+
+    @task
+    def validate_data(raw_data: bytes, **kwargs) -> bytes:
+        df = pickle.loads(raw_data)
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+        ctx = _get_context(dag_conf.get('store_mode', 'hotel'))
+        validated = ValidateDataStep(ctx).transform(df)
+        report = ctx.get_cached('validation_report') or {}
+        logging.info(f"Validation: {report.get('status')}, {report.get('sessions', 0)} sessions")
+        return pickle.dumps(validated)
+
+    @task
+    def extract_session_features(validated_data: bytes, **kwargs) -> bytes:
+        df = pickle.loads(validated_data)
+        if df.empty:
+            logging.warning("Empty input, skipping feature extraction")
+            return pickle.dumps(pd.DataFrame())
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+        ctx = _get_context(dag_conf.get('store_mode', 'hotel'))
+        features = ExtractSessionFeaturesStep(ctx).transform(df)
+        logging.info(f"Extracted {len(features.columns)} features for {len(features)} sessions")
+        return pickle.dumps(features)
+
+    @task
+    def join_labels(features_data: bytes, **kwargs) -> bytes:
+        features_df = pickle.loads(features_data)
+        if features_df.empty:
+            logging.warning("Empty features, skipping label join")
+            return pickle.dumps(pd.DataFrame())
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+        ctx = _get_context(dag_conf.get('store_mode', 'hotel'))
+        labeled = JoinLabelsStep(ctx).transform(features_df)
+        n_agents = labeled['is_agent'].sum() if 'is_agent' in labeled.columns else 0
+        logging.info(f"Labeled {len(labeled)} sessions: {n_agents} agents")
+        return pickle.dumps(labeled)
+
+    @task(outlets=[TRAINING_DATASET])
+    def publish_training_data(labeled_data: bytes, **kwargs) -> dict:
+        labeled_df = pickle.loads(labeled_data)
+        if labeled_df.empty:
+            return {'status': 'skipped', 'reason': 'empty_data'}
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+        return {
+            'status': 'success',
+            'n_sessions': len(labeled_df),
+            'n_features': len([c for c in labeled_df.columns if c not in ['sessionId', 'experimentId', 'is_agent']]),
+            'store_mode': dag_conf.get('store_mode', 'hotel'),
+            'timestamp': pd.Timestamp.now().isoformat(),
+        }
+
+    raw = fetch_interactions()
+    validated = validate_data(raw)
+    features = extract_session_features(validated)
+    labeled = join_labels(features)
+    publish_training_data(labeled)

experiments/airflow/dags/surge_pricing_factory.py

@@ -0,0 +1,220 @@
+from pandas.core.algorithms import factorize_array
+from airflow import DAG
+from airflow.operators.python import PythonOperator
+from airflow.utils.dates import days_ago
+from datetime import timedelta
+import pandas as pd
+import logging
+import sys
+import pickle
+
+sys.path.insert(0, '/opt/airflow')
+
+from procesing.context import PipelineContext
+from procesing.providers import SupabaseProvider, BackendAPIProvider
+from procesing.steps import (
+    FetchInteractionsStep,
+    FetchPriceLogsStep,
+    ComputeDemandStep,
+    AggregatePriceLogsStep,
+    JoinProductFeaturesStep,
+)
+from procesing.pricers.simple import SimpleSurgePricer
+
+DEFAULT_ARGS = {
+    'owner': 'phantom-research',
+    'depends_on_past': False,
+    'email_on_failure': False,
+    'email_on_retry': False,
+    'retries': 2,
+    'retry_delay': timedelta(minutes=5),
+}
+
+class CompositeProvider(SupabaseProvider, BackendAPIProvider):
+    def __init__(self):
+        SupabaseProvider.__init__(self)
+        BackendAPIProvider.__init__(self)
+
+def _get_provider():
+    return CompositeProvider()
+
+def _make_task_callables(store_mode: str):
+    """Generate task callables bound to a specific store_mode."""
+
+    def get_context(**kwargs):
+        return PipelineContext(provider=_get_provider(), store_mode=store_mode)
+
+    def fetch_interactions(**kwargs):
+        ctx = get_context(**kwargs)
+        df = FetchInteractionsStep(ctx).transform(None)
+        kwargs['ti'].xcom_push(key='interactions_raw', value=pickle.dumps(df))
+        logging.info(f"[{store_mode}] Fetched {len(df)} interaction records")
+        return len(df)
+
+    def fetch_price_logs(**kwargs):
+        ctx = get_context(**kwargs)
+        df = FetchPriceLogsStep(ctx).transform(None)
+        kwargs['ti'].xcom_push(key='price_logs_raw', value=pickle.dumps(df))
+        logging.info(f"[{store_mode}] Fetched {len(df)} price records")
+        return len(df)
+
+    def compute_demand(**kwargs):
+        ti = kwargs['ti']
+        df = pickle.loads(ti.xcom_pull(key='interactions_raw'))
+        ctx = get_context(**kwargs)
+        demand_df = ComputeDemandStep(ctx).transform(df)
+        ti.xcom_push(key='demand_data', value=pickle.dumps(demand_df))
+        logging.info(f"[{store_mode}] Computed demand for {len(demand_df)} products")
+        return len(demand_df)
+
+    def aggregate_price_logs(**kwargs):
+        ti = kwargs['ti']
+        df = pickle.loads(ti.xcom_pull(key='price_logs_raw'))
+        ctx = get_context(**kwargs)
+        price_df = AggregatePriceLogsStep(ctx).transform(df)
+        ti.xcom_push(key='price_data', value=pickle.dumps(price_df))
+        logging.info(f"[{store_mode}] Aggregated price logs for {len(price_df)} products")
+        return len(price_df)
+
+    def join_product_features(**kwargs):
+        ti = kwargs['ti']
+        demand_df = pickle.loads(ti.xcom_pull(key='demand_data'))
+        price_df = pickle.loads(ti.xcom_pull(key='price_data'))
+        ctx = get_context(**kwargs)
+        joined_df = JoinProductFeaturesStep(ctx).transform((demand_df, price_df))
+        ti.xcom_push(key='product_features', value=pickle.dumps(joined_df))
+        logging.info(f"[{store_mode}] Joined features for {len(joined_df)} products")
+        return len(joined_df)
+
+    def apply_surge_pricing(**kwargs):
+        ti = kwargs['ti']
+        product_features = pickle.loads(ti.xcom_pull(key='product_features'))
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+
+        data = product_features.rename(columns={'demand_score': 'demand'})
+        surge_pricer = SimpleSurgePricer(
+            high_threshold=dag_conf.get('high_threshold', 10),
+            low_threshold=dag_conf.get('low_threshold', 2),
+            surge_multiplier=dag_conf.get('surge_multiplier', 1.2),
+            discount_multiplier=dag_conf.get('discount_multiplier', 0.9)
+        )
+        surge_pricer.fit(data)
+        data['optimal_price'] = surge_pricer.predict()
+
+        prices_df = data[['productId', 'price', 'base_price', 'optimal_price', 'demand']].rename(columns={
+            'price': 'current_price', 'demand': 'demand_score'
+        })
+        ti.xcom_push(key='predicted_prices', value=pickle.dumps(prices_df))
+        logging.info(f"[{store_mode}] Applied surge pricing for {len(prices_df)} products")
+        return len(prices_df)
+
+    def publish_results(**kwargs):
+        ti = kwargs['ti']
+        prices_df = pickle.loads(ti.xcom_pull(key='predicted_prices'))
+        from lib.model_registry import ModelRegistry
+
+        registry = ModelRegistry()
+        dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+
+        metadata = {
+            'timestamp': pd.Timestamp.now().isoformat(),
+            'store_mode': store_mode,
+            'dag_run_id': kwargs['dag_run'].run_id if kwargs.get('dag_run') else 'manual',
+            'pricing_method': 'surge',
+            'high_threshold': dag_conf.get('high_threshold', 10),
+            'low_threshold': dag_conf.get('low_threshold', 2),
+            'surge_multiplier': dag_conf.get('surge_multiplier', 1.2),
+            'discount_multiplier': dag_conf.get('discount_multiplier', 0.9)
+        }
+        registry.publish_prices(prices_df, model_name=f'{store_mode}_latest', metadata=metadata)
+        logging.info(f"[{store_mode}] Published surge pricing for {len(prices_df)} products")
+
+        return {
+            'n_products': len(prices_df),
+            'registry_status': 'success',
+            'store_mode': store_mode,
+            'mean_demand': float(prices_df['demand_score'].mean()) if 'demand_score' in prices_df.columns else None
+        }
+
+    return {
+        'fetch_interactions': fetch_interactions,
+        'fetch_price_logs': fetch_price_logs,
+        'compute_demand': compute_demand,
+        'aggregate_price_logs': aggregate_price_logs,
+        'join_product_features': join_product_features,
+        'apply_surge_pricing': apply_surge_pricing,
+        'publish_results': publish_results,
+    }
+
+
+def create_surge_pricing_dag(store_mode: str) -> DAG:
+    """Factory: generates a surge pricing DAG for a given store_mode."""
+    callables = _make_task_callables(store_mode)
+
+    dag = DAG(
+        f'surge_pricing_{store_mode}',
+        default_args=DEFAULT_ARGS,
+        description=f'Surge pricing pipeline for {store_mode} store mode',
+        schedule_interval='*/15 * * * *',
+        start_date=days_ago(1),
+        catchup=False,
+        max_active_runs=1,
+        tags=['pricing', 'surge', 'research', store_mode],
+    )
+
+    with dag:
+        t_fetch_interactions = PythonOperator(
+            task_id='fetch_interactions',
+            python_callable=callables['fetch_interactions'],
+            provide_context=True,
+        )
+        t_fetch_price_logs = PythonOperator(
+            task_id='fetch_price_logs',
+            python_callable=callables['fetch_price_logs'],
+            provide_context=True,
+        )
+        t_compute_demand = PythonOperator(
+            task_id='compute_demand',
+            python_callable=callables['compute_demand'],
+            provide_context=True,
+        )
+        t_aggregate_prices = PythonOperator(
+            task_id='aggregate_price_logs',
+            python_callable=callables['aggregate_price_logs'],
+            provide_context=True,
+        )
+        t_join_features = PythonOperator(
+            task_id='join_product_features',
+            python_callable=callables['join_product_features'],
+            provide_context=True,
+        )
+        t_surge_pricing = PythonOperator(
+            task_id='apply_surge_pricing',
+            python_callable=callables['apply_surge_pricing'],
+            provide_context=True,
+        )
+        t_publish = PythonOperator(
+            task_id='publish_results',
+            python_callable=callables['publish_results'],
+            provide_context=True,
+        )
+
+        t_fetch_interactions >> t_compute_demand
+        t_fetch_price_logs >> t_aggregate_prices
+        [t_compute_demand, t_aggregate_prices] >> t_join_features >> t_surge_pricing >> t_publish
+
+    return dag
+
+
+# instantiate DAGs for Airflow to discover
+dag_airline = create_surge_pricing_dag('airline')
+dag_hotel = create_surge_pricing_dag('hotel')
+
+# TODO: Refactor this factory from a surge pricing factory to a general pricing factory
+# We will do this by passing a pricing strategy class to the factory, since the generic pipeline is:
+# take all interaction data, group by sessionId and assign a new price vector to each session
+# in the grouping we get a subset of the interactions per sessionId and we can map that to some Features
+# we define a custom _get_features(interactions .) methodin the strategy class
+# we then run only the inference which is the .predict(trajectory) per-session which will give us a new price vector
+# this we then publish for each sessionId group
+# this might include no deleting most of the pricers we have defined and starting with a super simple surge-pricing algorithm that is no-fit only predict. This we can then test end-to-end and observe changes to prices according to a desired strategy - we have to define this one as a very short term strategy because we run sessions that take only a few minutes.

experiments/airflow/dags/surge_pricing_pipeline.py

@@ -0,0 +1,253 @@
+from airflow import DAG
+from airflow.operators.python import PythonOperator
+from airflow.utils.dates import days_ago
+from datetime import timedelta
+import pandas as pd
+import logging
+import sys
+import pickle
+import io
+
+# add parent dir to path so procesing package can be imported
+sys.path.insert(0, '/opt/airflow')
+
+from procesing.context import PipelineContext
+from procesing.providers import SupabaseProvider, BackendAPIProvider
+from procesing.steps import (
+    FetchInteractionsStep,
+    FetchPriceLogsStep,
+    ComputeDemandStep,
+    AggregatePriceLogsStep,
+    JoinProductFeaturesStep,
+)
+from procesing.pricers.simple import SimpleSurgePricer
+
+default_args = {
+    'owner': 'phantom-research',
+    'depends_on_past': False,
+    'email_on_failure': False,
+    'email_on_retry': False,
+    'retries': 2,
+    'retry_delay': timedelta(minutes=5),
+}
+
+def get_provider():
+    """Factory to create composite provider"""
+    class CompositeProvider(SupabaseProvider, BackendAPIProvider): # TODO: Fix this into one global provider singelton instead of multiple inheritance declarations acoss the codebase
+        def __init__(self):
+            SupabaseProvider.__init__(self)
+            BackendAPIProvider.__init__(self)
+    return CompositeProvider()
+
+def get_context(**kwargs):
+    """Build pipeline context from Airflow config"""
+    dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+    return PipelineContext(
+        provider=get_provider(),
+        store_mode=dag_conf.get('store_mode', 'hotel'),
+    )
+
+# atomic task functions (each wraps one sklearn step)
+def fetch_interactions(**kwargs):
+    """Task: Fetch interaction data from Kafka"""
+    context = get_context(**kwargs)
+    step = FetchInteractionsStep(context)
+    df = step.transform(None)
+
+    kwargs['ti'].xcom_push(key='interactions_raw', value=pickle.dumps(df))
+    logging.info(f"Fetched {len(df)} interaction records")
+    return len(df)
+
+def fetch_price_logs(**kwargs):
+    """Task: Fetch price logs from Kafka"""
+    context = get_context(**kwargs)
+    step = FetchPriceLogsStep(context)
+    df = step.transform(None)
+
+    kwargs['ti'].xcom_push(key='price_logs_raw', value=pickle.dumps(df))
+    logging.info(f"Fetched {len(df)} price records")
+    return len(df)
+
+def compute_demand(**kwargs):
+    """Task: Compute demand scores from interactions"""
+    ti = kwargs['ti']
+    df = pickle.loads(ti.xcom_pull(key='interactions_raw'))
+
+    context = get_context(**kwargs)
+    step = ComputeDemandStep(context)
+    demand_df = step.transform(df)
+    # TODO: clear the xcom
+
+
+    ti.xcom_push(key='demand_data', value=pickle.dumps(demand_df))
+    logging.info(f"Computed demand for {len(demand_df)} products")
+    return len(demand_df)
+
+def aggregate_price_logs(**kwargs):
+    """Task: Aggregate price logs"""
+    ti = kwargs['ti']
+    df = pickle.loads(ti.xcom_pull(key='price_logs_raw'))
+
+    context = get_context(**kwargs)
+    step = AggregatePriceLogsStep(context)
+    price_df = step.transform(df)
+
+    ti.xcom_push(key='price_data', value=pickle.dumps(price_df))
+    logging.info(f"Aggregated price logs for {len(price_df)} products")
+    return len(price_df)
+
+def join_product_features(**kwargs):
+    """Task: Join demand and price data"""
+    ti = kwargs['ti']
+    demand_df = pickle.loads(ti.xcom_pull(key='demand_data'))
+    price_df = pickle.loads(ti.xcom_pull(key='price_data'))
+
+    context = get_context(**kwargs)
+    step = JoinProductFeaturesStep(context)
+    joined_df = step.transform((demand_df, price_df))
+
+    ti.xcom_push(key='product_features', value=pickle.dumps(joined_df))
+    logging.info(f"Joined features for {len(joined_df)} products")
+    return len(joined_df)
+
+def apply_surge_pricing(**kwargs):
+    """Task: Apply surge pricing rules to generate optimal prices"""
+    ti = kwargs['ti']
+    product_features = pickle.loads(ti.xcom_pull(key='product_features'))
+
+    dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+
+    # rename demand_score to demand for pricer compatibility
+    data = product_features.rename(columns={'demand_score': 'demand'})
+
+    high_thresh = dag_conf.get('high_threshold', 10)
+    low_thresh = dag_conf.get('low_threshold', 2)
+    surge_mult = dag_conf.get('surge_multiplier', 1.2)
+    discount_mult = dag_conf.get('discount_multiplier', 0.9)
+
+    logging.info(f"Surge pricing config: high_thresh={high_thresh}, low_thresh={low_thresh}, surge_mult={surge_mult}, discount_mult={discount_mult}")
+    logging.info(f"Demand stats: min={data['demand'].min():.2f}, max={data['demand'].max():.2f}, mean={data['demand'].mean():.2f}")
+    logging.info(f"Products with high demand (>={high_thresh}): {(data['demand'] >= high_thresh).sum()}")
+    logging.info(f"Products with low demand (<={low_thresh}): {(data['demand'] <= low_thresh).sum()}")
+
+    surge_pricer = SimpleSurgePricer(
+        high_threshold=high_thresh,
+        low_threshold=low_thresh,
+        surge_multiplier=surge_mult,
+        discount_multiplier=discount_mult
+    )
+    surge_pricer.fit(data)
+    data['optimal_price'] = surge_pricer.predict()
+
+    base_avg = data['base_price'].mean()
+    optimal_avg = data['optimal_price'].mean()
+    price_change_pct = ((optimal_avg - base_avg) / base_avg) * 100
+
+    logging.info(f"Price adjustment: base_avg={base_avg:.2f}, optimal_avg={optimal_avg:.2f}, change={price_change_pct:+.1f}%")
+
+    prices_df = data[['productId', 'price', 'base_price', 'optimal_price', 'demand']].rename(columns={
+        'price': 'current_price',
+        'demand': 'demand_score'
+    })
+
+    ti.xcom_push(key='predicted_prices', value=pickle.dumps(prices_df))
+    logging.info(f"Applied surge pricing for {len(prices_df)} products")
+    return len(prices_df)
+
+def publish_results(**kwargs):
+    """Task: Publish surge pricing results to registry"""
+    ti = kwargs['ti']
+    prices_df = pickle.loads(ti.xcom_pull(key='predicted_prices'))
+
+    sys.path.insert(0, '/opt/airflow')
+    from lib.model_registry import ModelRegistry
+
+    registry = ModelRegistry()
+    dag_conf = kwargs.get('dag_run').conf if kwargs.get('dag_run') else {}
+
+    metadata = {
+        'timestamp': pd.Timestamp.now().isoformat(),
+        'store_mode': dag_conf.get('store_mode', 'hotel'),
+        'dag_run_id': kwargs['dag_run'].run_id if kwargs.get('dag_run') else 'manual',
+        'pricing_method': 'surge',
+        'high_threshold': dag_conf.get('high_threshold', 10),
+        'low_threshold': dag_conf.get('low_threshold', 2),
+        'surge_multiplier': dag_conf.get('surge_multiplier', 1.2),
+        'discount_multiplier': dag_conf.get('discount_multiplier', 0.9)
+    }
+
+    registry.publish_prices(prices_df, model_name='latest', metadata=metadata)
+
+    logging.info(f"Published surge pricing for {len(prices_df)} products")
+
+    return {
+        'n_products': len(prices_df),
+        'registry_status': 'success',
+        'mean_demand': float(prices_df['demand_score'].mean()) if 'demand_score' in prices_df.columns else None
+    }
+
+
+# DAG definition
+with DAG(
+    'surge_pricing_pipeline',
+    default_args=default_args,
+    description='Simple surge pricing pipeline: demand aggregation + rule-based pricing',
+    schedule_interval='*/15 * * * *',
+    start_date=days_ago(1),
+    catchup=False,
+    max_active_runs=1,
+    tags=['pricing', 'surge', 'research', 'simplified'],
+) as dag:
+
+    # parallel data fetching
+    t_fetch_interactions = PythonOperator(
+        task_id='fetch_interactions',
+        python_callable=fetch_interactions,
+        provide_context=True,
+    )
+
+    t_fetch_price_logs = PythonOperator(
+        task_id='fetch_price_logs',
+        python_callable=fetch_price_logs,
+        provide_context=True,
+    )
+
+    # compute demand from interactions
+    t_compute_demand = PythonOperator(
+        task_id='compute_demand',
+        python_callable=compute_demand,
+        provide_context=True,
+    )
+
+    # aggregate price logs
+    t_aggregate_prices = PythonOperator(
+        task_id='aggregate_price_logs',
+        python_callable=aggregate_price_logs,
+        provide_context=True,
+    )
+
+    # join demand and prices
+    t_join_features = PythonOperator(
+        task_id='join_product_features',
+        python_callable=join_product_features,
+        provide_context=True,
+    )
+
+    # apply surge pricing
+    t_surge_pricing = PythonOperator(
+        task_id='apply_surge_pricing',
+        python_callable=apply_surge_pricing,
+        provide_context=True,
+    )
+
+    # publish to registry
+    t_publish = PythonOperator(
+        task_id='publish_results',
+        python_callable=publish_results,
+        provide_context=True,
+    )
+
+    # dependency graph: parallel fetch -> process -> join -> surge -> publish
+    t_fetch_interactions >> t_compute_demand
+    t_fetch_price_logs >> t_aggregate_prices
+    [t_compute_demand, t_aggregate_prices] >> t_join_features >> t_surge_pricing >> t_publish

experiments/ml/__init__.py

@@ -0,0 +1,11 @@
+from .evals import evaluate
+from .arch import (
+    XGBoostAgentClassifier,
+    LightGBMAgentClassifier
+)
+
+__all__ =[
+    'evaluate',
+    'XGBoostAgentClassifier',
+    'LightGBMAgentClassifier'
+]

experiments/ml/arch.py

@@ -0,0 +1,122 @@
+# sklearn compatible models for agent detection
+from sklearn.base import BaseEstimator, ClassifierMixin
+from procesing.context import PipelineContext
+from typing import Any, Optional, Tuple
+from abc import ABC, abstractmethod
+import xgboost as xgb
+import lightgbm as lgb
+import numpy as np
+import pandas as pd
+
+TASK = 'classification'
+LABELS = ['human', 'agent']
+
+
+class BaseAgentClassifier(BaseEstimator, ClassifierMixin, ABC):
+    """Base class for tree-based agent detection classifiers with common logic"""
+
+    def __init__(self, context: Optional[PipelineContext] = None, n_estimators: int = 200,
+                 max_depth: int = 6, learning_rate: float = 0.05,
+                 early_stopping_rounds: int = 20):
+        self.context = context
+        self.n_estimators = n_estimators
+        self.max_depth = max_depth
+        self.learning_rate = learning_rate
+        self.early_stopping_rounds = early_stopping_rounds
+        self.model_ = None
+        self.feature_names_ = None
+
+    def _to_array(self, X):
+        """Convert pandas structures to numpy arrays"""
+        return X.values if isinstance(X, (pd.DataFrame, pd.Series)) else X
+
+    def _compute_pos_weight(self, y_arr):
+        """Calculate scale_pos_weight for class imbalance handling"""
+        n_neg, n_pos = (y_arr == 0).sum(), (y_arr == 1).sum()
+        return n_neg / n_pos if n_pos > 0 else 1.0
+
+    def _prepare_eval_set(self, eval_set):
+        """Convert eval_set to numpy arrays if needed"""
+        if not eval_set:
+            return None
+        X_val, y_val = eval_set[0]
+        return [(self._to_array(X_val), self._to_array(y_val))]
+
+    @abstractmethod
+    def _build_model(self, scale_pos: float):
+        """Build the underlying model instance (must be implemented by subclasses)"""
+        pass
+
+    @abstractmethod
+    def _fit_with_eval(self, X_arr, y_arr, eval_arr):
+        """Fit model with evaluation set (must be implemented by subclasses)"""
+        pass
+
+    def fit(self, X, y, eval_set=None):
+        X_arr, y_arr = self._to_array(X), self._to_array(y)
+
+        if isinstance(X, pd.DataFrame):
+            self.feature_names_ = X.columns.tolist()
+
+        scale_pos = self._compute_pos_weight(y_arr)
+        self.model_ = self._build_model(scale_pos)
+
+        eval_arr = self._prepare_eval_set(eval_set)
+        if eval_arr:
+            self._fit_with_eval(X_arr, y_arr, eval_arr)
+        else:
+            self.model_.fit(X_arr, y_arr)
+
+        return self
+
+    def predict(self, X):
+        return self.model_.predict(self._to_array(X))
+
+    def predict_proba(self, X):
+        return self.model_.predict_proba(self._to_array(X))
+
+    @property
+    def feature_importances_(self):
+        return self.model_.feature_importances_ if self.model_ else None
+
+
+class XGBoostAgentClassifier(BaseAgentClassifier):
+    """XGBoost binary classifier for agent detection with class imbalance handling"""
+
+    def _build_model(self, scale_pos: float):
+        return xgb.XGBClassifier(
+            n_estimators=self.n_estimators,
+            max_depth=self.max_depth,
+            learning_rate=self.learning_rate,
+            scale_pos_weight=scale_pos,
+            eval_metric='auc',
+            early_stopping_rounds=self.early_stopping_rounds,
+            random_state=42,
+            tree_method='hist',
+            enable_categorical=False
+        )
+
+    def _fit_with_eval(self, X_arr, y_arr, eval_arr):
+        self.model_.fit(X_arr, y_arr, eval_set=eval_arr, verbose=False)
+
+
+class LightGBMAgentClassifier(BaseAgentClassifier):
+    """LightGBM binary classifier for agent detection with class imbalance handling"""
+
+    def _build_model(self, scale_pos: float):
+        return lgb.LGBMClassifier(
+            n_estimators=self.n_estimators,
+            max_depth=self.max_depth,
+            learning_rate=self.learning_rate,
+            scale_pos_weight=scale_pos,
+            metric='auc',
+            random_state=42,
+            verbosity=-1
+        )
+
+    def _fit_with_eval(self, X_arr, y_arr, eval_arr):
+        self.model_.fit(
+            X_arr, y_arr,
+            eval_set=eval_arr,
+            callbacks=[lgb.early_stopping(self.early_stopping_rounds, verbose=False)]
+        )

experiments/ml/evals.py

@@ -0,0 +1,103 @@
+from sklearn.metrics import (accuracy_score, precision_score, recall_score,
+                             f1_score, roc_auc_score, confusion_matrix, roc_curve)
+from torch.utils.tensorboard import SummaryWriter
+from logging import getLogger
+import numpy as np
+import matplotlib.pyplot as plt
+import io
+from PIL import Image
+
+logger = getLogger(__name__)
+
+
+def log_feature_importance(writer, model, feature_names, epoch):
+    """Visualize and log feature importance to TensorBoard"""
+    if not hasattr(model, 'feature_importances_') or model.feature_importances_ is None:
+        return
+
+    importance = model.feature_importances_
+    indices = np.argsort(importance)[::-1][:20]  # top 20
+    top_features = [feature_names[i] for i in indices]
+    top_importance = importance[indices]
+
+    for i, (feat, imp) in enumerate(zip(top_features, top_importance)):
+        writer.add_scalar(f'FeatureImportance/{feat}', imp, epoch)
+
+    fig, ax = plt.subplots(figsize=(10, 8))
+    ax.barh(range(len(top_features)), top_importance, align='center')
+    ax.set_yticks(range(len(top_features)))
+    ax.set_yticklabels(top_features)
+    ax.invert_yaxis()
+    ax.set_xlabel('Importance')
+    ax.set_title(f'Top 20 Feature Importance (Epoch {epoch})')
+    ax.grid(axis='x', alpha=0.3)
+
+    buf = io.BytesIO()
+    plt.tight_layout()
+    plt.savefig(buf, format='png', dpi=100)
+    buf.seek(0)
+    img = Image.open(buf)
+    img_arr = np.array(img)
+    writer.add_image('FeatureImportance/Chart', img_arr, epoch, dataformats='HWC')
+    plt.close()
+
+def evaluate(perdicted_class, predicted_proba, true_class, writer: SummaryWriter, epoch: int):
+    accuracy = accuracy_score(true_class, perdicted_class)
+    precision = precision_score(true_class, perdicted_class, zero_division=0)
+    recall = recall_score(true_class, perdicted_class, zero_division=0)
+    f1 = f1_score(true_class, perdicted_class, zero_division=0)
+    roc_auc = roc_auc_score(true_class, predicted_proba)
+
+    writer.add_scalar('Eval/Accuracy', accuracy, epoch)
+    writer.add_scalar('Eval/Precision', precision, epoch)
+    writer.add_scalar('Eval/Recall', recall, epoch)
+    writer.add_scalar('Eval/F1_Score', f1, epoch)
+    writer.add_scalar('Eval/ROC_AUC', roc_auc, epoch)
+
+    # confusion matrix
+    cm = confusion_matrix(true_class, perdicted_class)
+    tn, fp, fn, tp = cm.ravel()
+    writer.add_scalar('Eval/TrueNeg', tn, epoch)
+    writer.add_scalar('Eval/FalsePos', fp, epoch)
+    writer.add_scalar('Eval/FalseNeg', fn, epoch)
+    writer.add_scalar('Eval/TruePos', tp, epoch)
+
+    # specificity and sensitivity
+    specificity = tn / (tn + fp) if (tn + fp) > 0 else 0
+    sensitivity = recall  # same as recall/TPR
+    writer.add_scalar('Eval/Specificity', specificity, epoch)
+    writer.add_scalar('Eval/Sensitivity', sensitivity, epoch)
+
+    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 4))
+    ax1.matshow(cm, cmap='Blues', alpha=0.7)
+    for i in range(2):
+        for j in range(2):
+            ax1.text(j, i, str(cm[i, j]), ha='center', va='center', fontsize=14)
+    ax1.set_xlabel('Predicted')
+    ax1.set_ylabel('True')
+    ax1.set_title(f'Confusion Matrix (Epoch {epoch})')
+    ax1.set_xticks([0, 1])
+    ax1.set_yticks([0, 1])
+    ax1.set_xticklabels(['Human', 'Agent'])
+    ax1.set_yticklabels(['Human', 'Agent'])
+
+    # ROC curve
+    fpr, tpr, _ = roc_curve(true_class, predicted_proba)
+    ax2.plot(fpr, tpr, label=f'AUC={roc_auc:.3f}', linewidth=2)
+    ax2.plot([0, 1], [0, 1], 'k--', label='Random')
+    ax2.set_xlabel('False Positive Rate')
+    ax2.set_ylabel('True Positive Rate')
+    ax2.set_title('ROC Curve')
+    ax2.legend()
+    ax2.grid(alpha=0.3)
+
+    buf = io.BytesIO()
+    plt.tight_layout()
+    plt.savefig(buf, format='png', dpi=100)
+    buf.seek(0)
+    img = Image.open(buf)
+    img_arr = np.array(img)
+    writer.add_image('Eval/Metrics', img_arr, epoch, dataformats='HWC')
+    plt.close()
+
+    logger.info(f"Eval {epoch}: Acc={accuracy:.4f} Prec={precision:.4f} Rec={recall:.4f} F1={f1:.4f} AUC={roc_auc:.4f}")

experiments/ml/requirements.txt

@@ -0,0 +1,6 @@
+torch
+tensorboard
+fastparquet
+pyarrow
+xgboost
+lightgbm

experiments/ml/train.py

@@ -0,0 +1,137 @@
+from torch.utils.tensorboard import SummaryWriter
+from sklearn.model_selection import train_test_split
+from logging import getLogger
+from pathlib import Path
+import pandas as pd
+import numpy as np
+import joblib
+from datetime import datetime
+from ml.evals import evaluate, log_feature_importance
+from ml.arch import XGBoostAgentClassifier, LightGBMAgentClassifier, LABELS
+
+logger = getLogger(__name__)
+
+FEATURE_COLS_EXCLUDE = ['sessionId', 'experimentId', 'is_agent', 'xp_human_only', 'xp_market_mode', 'browser_family']
+RUNS_DIR = Path('ml/runs')
+CHECKPOINTS_DIR = Path('ml/checkpoints')
+
+
+def prepare_data(df):
+    """
+    Prepare feature matrix and labels from raw dataframe
+    Handles missing labels, feature selection, and categorical encoding
+    Returns: (X, y, feature_cols)
+    """
+    # drop rows with missing labels
+    n_before = len(df)
+    df = df[df['is_agent'].notna()].copy()
+    n_dropped = n_before - len(df)
+    if n_dropped > 0:
+        logger.warning(f"Dropped {n_dropped} sessions with missing labels")
+
+    if len(df) == 0:
+        logger.error("No labeled data available")
+        return None, None, None
+
+    feature_cols = [c for c in df.columns if c not in FEATURE_COLS_EXCLUDE]
+
+    # handle categorical browser_family via one-hot encoding
+    if 'browser_family' in df.columns:
+        browser_dummies = pd.get_dummies(df['browser_family'], prefix='browser', drop_first=True)
+        df = pd.concat([df, browser_dummies], axis=1)
+        feature_cols.extend(browser_dummies.columns.tolist())
+
+    X = df[feature_cols].fillna(0)
+    y = df['is_agent'].astype(int)
+
+    return X, y, feature_cols
+
+
+def train(data_path=None, model_type='xgboost', test_size=0.2, random_state=42,
+          n_estimators=200, max_depth=6, learning_rate=0.05):
+    """
+    Train agent detection classifier
+    Args:
+        data_path: path to labeled feature matrix CSV or parquet
+        model_type: 'xgboost' or 'lightgbm'
+        test_size: fraction for test split
+        random_state: seed for reproducibility
+    """
+    RUNS_DIR.mkdir(exist_ok=True)
+    CHECKPOINTS_DIR.mkdir(exist_ok=True)
+
+    run_name = f"{model_type}_{datetime.now().strftime('%Y%m%d_%H%M%S')}"
+    writer = SummaryWriter(log_dir=RUNS_DIR / run_name)
+    logger.info(f"Starting training run: {run_name}")
+
+    # load data
+    if data_path is None:
+        logger.error("data_path required")
+        return
+    df = pd.read_parquet(data_path)
+    logger.info(f"Loaded {len(df)} sessions from {data_path}")
+
+    # prepare features and labels
+    if 'is_agent' not in df.columns:
+        logger.error("Missing is_agent column")
+        return
+
+    X, y, feature_cols = prepare_data(df)
+    if X is None:
+        return
+
+    # class distribution
+    n_agents = y.sum()
+    n_humans = (y == 0).sum()
+    logger.info(f"Class distribution: {n_humans} humans, {n_agents} agents" + (f" (ratio {n_humans / n_agents:.2f})" if n_agents > 0 else ""))
+
+    # train/test split with stratification
+    X_train, X_test, y_train, y_test = train_test_split(
+        X, y, test_size=test_size, random_state=random_state, stratify=y
+    )
+    logger.info(f"Train: {len(X_train)}, Test: {len(X_test)}")
+
+    # init model
+    if model_type == 'xgboost':
+        model = XGBoostAgentClassifier(
+            n_estimators=n_estimators,
+            max_depth=max_depth,
+            learning_rate=learning_rate
+        )
+    elif model_type == 'lightgbm':
+        model = LightGBMAgentClassifier(
+            n_estimators=n_estimators,
+            max_depth=max_depth,
+            learning_rate=learning_rate
+        )
+    else:
+        logger.error(f"Unknown model type: {model_type}")
+        return
+
+    # train with eval set for early stopping
+    model.fit(X_train, y_train, eval_set=[(X_test, y_test)])
+    logger.info("Training complete")
+
+    # evaluate on test set
+    y_pred = model.predict(X_test)
+    y_prob = model.predict_proba(X_test)[:, 1]
+
+    evaluate(y_pred, y_prob, y_test, writer, epoch=0)
+
+    # log feature importance
+    log_feature_importance(writer, model, X.columns.tolist(), epoch=0)
+
+    # save model
+    model_path = CHECKPOINTS_DIR / f"{run_name}.pkl"
+    joblib.dump({'model': model, 'feature_cols': X.columns.tolist(), 'run_name': run_name}, model_path)
+    logger.info(f"Model saved to {model_path}")
+
+    writer.close()
+    return model, X.columns.tolist()
+
+
+if __name__ == "__main__":
+    import sys
+    data_path = sys.argv[1]
+    model_type = sys.argv[2] if len(sys.argv) > 2 else 'xgboost'
+    train(data_path, model_type=model_type)

experiments/procesing/__init__.py

@@ -12,16 +12,14 @@ from procesing.steps import (
     ComputeDemandStep,
     ComputeDemandForChunksStep,
     AggregatePriceLogsStep,
-    ComputeElasticityStep,
+    # StateSpace,
-    StateSpace,
+    # BuildStateSpaceStep,
-    BuildStateSpaceStep,
     FitPricingFunctionStep,
     PredictPricesStep,
 )
 from procesing.pipelines import (
     interaction_extraction_pipeline,
     price_extraction_pipeline,
-    elasticity_computation_pipeline,
     pricing_pipeline,
     full_pipeline,
 )
@@ -42,14 +40,12 @@ __all__ = [
     'ComputeDemandStep',
     'ComputeDemandForChunksStep',
     'AggregatePriceLogsStep',
-    'ComputeElasticityStep',
+    # 'StateSpace',
-    'StateSpace',
+    # 'BuildStateSpaceStep',
-    'BuildStateSpaceStep',
     'FitPricingFunctionStep',
     'PredictPricesStep',
     'interaction_extraction_pipeline',
     'price_extraction_pipeline',
-    'elasticity_computation_pipeline',
     'pricing_pipeline',
     'full_pipeline',
 ]

experiments/procesing/pipelines.py

@@ -2,7 +2,7 @@ from sklearn.pipeline import Pipeline
 import pandas as pd
 from procesing.context import PipelineContext
 from procesing.providers import SupabaseProvider, BackendAPIProvider
-from typing import Union
+import os
 from procesing.steps import (
     FetchInteractionsStep,
     FetchPriceLogsStep,
@@ -13,11 +13,15 @@ from procesing.steps import (
     ChunkByTimeWindowStep,
     ComputeDemandForChunksStep,
     AggregatePriceLogsStep,
-    ComputeElasticityStep,
-    BuildStateSpaceStep,
     FitPricingFunctionStep,
     PredictPricesStep,
+    ComputeDemandStep,
+    JoinProductFeaturesStep,
+    ExtractSessionFeaturesStep,
+    JoinLabelsStep,
+    ValidateDataStep,
 )
+from procesing.pricers import SimpleSurgePricer
 
 def interaction_extraction_pipeline(context: PipelineContext):
     """Pipeline for extracting and augmenting interaction data"""
@@ -35,104 +39,136 @@ def price_extraction_pipeline(context: PipelineContext):
     ])
 
 
-def elasticity_computation_pipeline(context: PipelineContext,
+def product_features_pipeline(context: PipelineContext,
                                     interactions_df: pd.DataFrame,
                                     price_logs_df: pd.DataFrame):
-    """
+    demand_step = ComputeDemandStep(context)
-    Compute elasticity from interactions and price logs.
-    Manual orchestration needed for branching logic.
-    """
-    # branch 1: chunk interactions and compute demand
-    chunk_step = ChunkByTimeWindowStep(context)
-    interaction_chunks = chunk_step.transform(interactions_df)
-
-    demand_step = ComputeDemandForChunksStep(context)
-    demand_chunks = demand_step.transform(interaction_chunks)
-
-    # branch 2: aggregate price logs
     price_step = AggregatePriceLogsStep(context)
-    price_chunks = price_step.transform(price_logs_df)
+    join_step = JoinProductFeaturesStep(context)
-
-    # convergence: compute elasticity
-    elasticity_step = ComputeElasticityStep(context)
-    elasticity_df = elasticity_step.transform((demand_chunks, price_chunks))
-
-    return elasticity_df
 
 
-def pricing_pipeline(context: PipelineContext, elasticity_df: pd.DataFrame):
+    demand_data = demand_step.transform(interactions_df)
+    price_data= price_step.transform(price_logs_df)
+    joined_data = join_step.transform((demand_data, price_data))
+
+    return joined_data
+
+
+
+def pricing_pipeline(context: "PipelineContext",
+                     data: pd.DataFrame,
+                     high_threshold: int = 10,
+                     low_threshold: int = 2,
+                     surge_multiplier: float = 1.2,
+                     discount_multiplier: float = 0.9) -> pd.DataFrame:
+
+    if data.empty or 'productId' not in data.columns:
+        return pd.DataFrame()
+
+    surge_pricer = SimpleSurgePricer()
+    surge_pricer.fit(data)
+    data['optimal_price'] = surge_pricer.predict()
+    return data
+
+
+def full_pipeline(context: PipelineContext,
+                  high_threshold: int = 10,
+                  low_threshold: int = 2,
+                  surge_multiplier: float = 1.2,
+                  discount_multiplier: float = 0.9):
     """
-    Generate optimal prices from elasticity estimates.
+    Complete end-to-end pipeline: data extraction -> demand/price aggregation -> surge pricing
+
+    Args:
+        context: Pipeline context
+        high_threshold: Demand threshold for surge pricing
+        low_threshold: Demand threshold for discounts
+        surge_multiplier: Price multiplier for high demand
+        discount_multiplier: Price multiplier for low demand
+
+    Returns:
+        tuple: (product_features_df, optimal_prices_df)
+            - product_features_df: [productId, demand_score, price]
+            - optimal_prices_df: [productId, current_price, optimal_price, demand_score]
     """
-    # build state space
-    state_step = BuildStateSpaceStep(context)
-    state_space = state_step.transform(elasticity_df)
-
-    # fit pricing function
-    fit_step = FitPricingFunctionStep(context)
-    pricer = fit_step.transform(elasticity_df)
-
-    # predict prices
-    predict_step = PredictPricesStep(context)
-    prices_df = predict_step.transform((pricer, state_space))
-
-    return prices_df
-
-
-def full_pipeline(context: PipelineContext):
-    """
-    Complete end-to-end pipeline: data extraction -> elasticity -> pricing
-    Returns: (elasticity_df, prices_df)
-    """
-    # extract interactions
     interaction_pipe = interaction_extraction_pipeline(context)
-    interactions_df = interaction_pipe.fit_transform(None)
-
-    # extract price logs
     price_pipe = price_extraction_pipeline(context)
+
+    interactions_df = interaction_pipe.fit_transform(None)
     price_logs_df = price_pipe.fit_transform(None)
+    product_features_df = product_features_pipeline(context, interactions_df, price_logs_df)
+    print(product_features_df.to_string())
 
-    if interactions_df.empty or price_logs_df.empty:
+    # generate optimal prices using surge rules
-        return None, None
+    optimal_prices_df = pricing_pipeline(context, product_features_df,
+                                          high_threshold=high_threshold,
+                                          low_threshold=low_threshold,
+                                          surge_multiplier=surge_multiplier,
+                                          discount_multiplier=discount_multiplier)
 
-    # compute elasticity
+    return product_features_df, optimal_prices_df
-    elasticity_df = elasticity_computation_pipeline(
-        context,
-        interactions_df,
-        price_logs_df
-    )
 
-    if elasticity_df is None or elasticity_df.empty:
-        return elasticity_df, None
 
-    # generate prices
+def ml_training_pipeline(context: PipelineContext) -> pd.DataFrame:
-    prices_df = pricing_pipeline(context, elasticity_df)
+    """
+    Build labeled session-level feature matrix for ML model training.
+    Pipeline: fetch -> validate -> extract features -> join labels
+
+    Returns:
+        DataFrame with ~25 features per session + is_agent label
+        Columns: sessionId, experimentId, temporal/behavioral/product/ua features, is_agent
+    """
+    # fetch raw interactions
+    interactions_df = FetchInteractionsStep(context).transform(None)
+
+    # validate data quality (report cached in context)
+    interactions_df = ValidateDataStep(context).transform(interactions_df)
+    if interactions_df.empty:
+        return pd.DataFrame()
+
+    # extract vectorized session features
+    features_df = ExtractSessionFeaturesStep(context).transform(interactions_df)
+    if features_df.empty:
+        return pd.DataFrame()
+
+    # join experiment labels (is_agent = ~xp_human_only)
+    labeled_df = JoinLabelsStep(context).transform(features_df)
+
+    return labeled_df
+
 
-    return elasticity_df, prices_df
 
 
 if __name__ == '__main__':
 
-    class Provider(SupabaseProvider, BackendAPIProvider):
+    class ExperimentsProvider(SupabaseProvider, BackendAPIProvider):
-        def __init__(self, backend_url: str):
+        def fetch_kafka_topic(self, topic: str) -> pd.DataFrame:
-            SupabaseProvider.__init__(self)
+            base_path = "/home/velocitatem/Documents/Projects/PHANTOM/experiments/collected_data/" # os.path.join(os.path.dirname(__file__), "collected_data")
-            BackendAPIProvider.__init__(self, backend_url=backend_url)
+            if not os.path.isdir(base_path):
-    # example run
+                return pd.DataFrame()
-    context = PipelineContext(
-        provider=Provider(backend_url="http://localhost:5000"),
-        store_mode='hotel',
-    )
 
-    elasticity_df, prices_df = full_pipeline(context)
+            files = {"user-interactions": "int.json", "price-logs": "price.json"}
+            file_to_read = files.get(topic, files["user-interactions"])
+            frames = []
 
-    if elasticity_df is not None and not elasticity_df.empty:
+            for d in os.listdir(base_path):
-        print("Elasticity Estimates:")
+                full_path = os.path.join(base_path, d, file_to_read)
-        print(elasticity_df.to_string(index=False))
+                if not os.path.isfile(full_path):
-    else:
+                    continue
-        print("No elasticity estimates computed.")
+                try:
+                    data = pd.read_json(full_path)
+                    payloads = pd.DataFrame([r['payload'] for r in data['value'].to_list()])
+                    frames.append(payloads)
+                except Exception as e:
+                    print(f"Warning: Could not process {full_path}: {e}")
 
-    if prices_df is not None and not prices_df.empty:
+            return pd.concat(frames, ignore_index=True) if frames else pd.DataFrame()
-        print("\nPredicted Prices:")
+
-        print(prices_df.to_string(index=False))
+    # demo: run ML training pipeline
-    else:
+    context = PipelineContext(provider=ExperimentsProvider(), store_mode='hotel')
-        print("No prices predicted.")
+    features = ml_training_pipeline(context)
+    print(f"Feature matrix: {features.shape}")
+    print(features.head())
+    print(features.info())
+
+    features.to_parquet("features.parquet")

experiments/procesing/pricers/__init__.py

@@ -1,6 +1,6 @@
 from procesing.pricers.base import PricingFunction
 from procesing.pricers.elasticity import ElasticityBasedPricer
-from procesing.pricers.simple import StaticPricer, RandomPricer
+from procesing.pricers.simple import StaticPricer, RandomPricer, SimpleSurgePricer
 from procesing.pricers.session_aware import SessionAwarePricer, ProductSpecificSessionPricer
 
 __all__ = [
@@ -8,6 +8,7 @@ __all__ = [
     'ElasticityBasedPricer',
     'StaticPricer',
     'RandomPricer',
+    'SimpleSurgePricer',
     'SessionAwarePricer',
     'ProductSpecificSessionPricer'
 ]

experiments/procesing/pricers/base.py

@@ -7,15 +7,6 @@ import pandas as pd
 class PricingFunction(ABC):
     """
     Abstract base for pricing functions.
-
-    Defines mapping: f(Q_t, P_t, S_t, H_t) -> P_{t+1}
-
-    Where:
-        Q_t ∈ R^n: demand vector at time t
-        P_t ∈ R^n: price vector at time t
-        S_t: session features (behavioral signals, interactions)
-        H_t = {Q_{t-k}, P_{t-k}, S_{t-k}}: historical state trajectory
-
     Objective:
         maximize E[R_T] = E[Σ P_t^T · Q_t]
         subject to:
@@ -25,26 +16,32 @@ class PricingFunction(ABC):
     """
 
     @abstractmethod
-    def fit(self, historical_data: pd.DataFrame, **kwargs):
+    def fit(self, *kwargs):
         """
         Offline training on historical data.
+        This is where we can think about some maximization of expected revenue
+        over historical trajectories to learn parameters of the pricing function.
+        (This however we cover move in the RL side of things)
 
-        Args:
-            historical_data: DataFrame with elasticity, prices, demand signals
-            **kwargs: additional training parameters
         """
         pass
 
     @abstractmethod
-    def predict(self, state_space) -> np.ndarray:
+    def predict(self, *kwargs) -> np.ndarray:
         """
         Generate optimal prices given current state.
+        This is an abstract method that transitions from τ -> P*
+        which is the mapping from the trajectory to optimal prices under
+        some subset of session grouping (so, per sessionId)
+        """
+        pass
 
-        Args:
+    @abstractmethod
-            state_space: StateSpace object containing Q_t, P_t, S_t, H_t
+    def _get_features(self, *kwargs) -> np.ndarray:
-
+        """
+        Extract features from trajectory for pricing decision.
         Returns:
-            P_{t+1}: price vector in R^n
+            np.ndarray of shape (n_products, n_features)
         """
         pass
 

experiments/procesing/pricers/elasticity.py

@@ -57,3 +57,13 @@ class ElasticityBasedPricer(PricingFunction):
         # enforce bounds
         prices = np.clip(prices, self.price_floor, self.price_ceil)
         return prices
+
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Extract elasticity, demand, and demand deviation for each product"""
+        if state_space is None or self.elasticity is None:
+            n = len(self.elasticity) if self.elasticity is not None else 0
+            return np.zeros((n, 3))
+
+        demand = np.asarray(state_space.demand)
+        demand_dev = (demand - self.mean_demand) / (self.mean_demand + 1e-6)
+        return np.column_stack([self.elasticity, demand, demand_dev])

experiments/procesing/pricers/session_aware.py

@@ -107,6 +107,36 @@ class SessionAwarePricer(PricingFunction):
 
         return prices
 
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Extract elasticity, demand, and session features"""
+        if state_space is None or self.elasticity is None:
+            n = len(self.elasticity) if self.elasticity is not None else 0
+            return np.zeros((n, 5))
+
+        demand = np.asarray(state_space.demand)
+        n_products = len(demand)
+
+        # extract session features
+        velocity = 0.0
+        view_depth = 0.0
+        cart_to_view = 0.0
+
+        if not state_space.session_features.empty:
+            sf = state_space.session_features.iloc[0]
+            velocity = sf.get('interaction_velocity', 0.0)
+            view_depth = sf.get('product_view_depth', 0.0)
+            cart_to_view = sf.get('cart_to_view_ratio', 0.0)
+
+        # broadcast session features to all products
+        features = np.column_stack([
+            self.elasticity,
+            demand,
+            np.full(n_products, velocity),
+            np.full(n_products, view_depth),
+            np.full(n_products, cart_to_view)
+        ])
+        return features
+
 
 class ProductSpecificSessionPricer(PricingFunction):
     """
@@ -170,3 +200,12 @@ class ProductSpecificSessionPricer(PricingFunction):
 
         prices = np.clip(base_prices, self.price_floor, self.price_ceil)
         return prices
+
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Extract elasticity and demand features for product-specific pricing"""
+        if state_space is None or self.elasticity is None:
+            n = len(self.elasticity) if self.elasticity is not None else 0
+            return np.zeros((n, 2))
+
+        demand = np.asarray(state_space.demand)
+        return np.column_stack([self.elasticity, demand])

experiments/procesing/pricers/simple.py

@@ -3,6 +3,46 @@ import pandas as pd
 from procesing.pricers.base import PricingFunction
 
 
+def session_features_to_demand(session_features: pd.DataFrame) -> float:
+    """
+    Map session behavioral features to demand proxy.
+    THIS is the critical θ̂ → D transformation for rule-based pricing.
+
+    Logic:
+      - High velocity → agent behavior → price up (revenue recovery)
+      - High cart ratio → purchase intent → price up
+      - Low activity → discount to convert
+
+    Returns: demand proxy score (0-20 range, higher = more demand)
+    """
+    if session_features.empty:
+        return 1.0
+
+    feat = session_features.iloc[0] if len(session_features) > 0 else {}
+
+    velocity = feat.get('interaction_velocity', 0)
+    cart_ratio = feat.get('cart_to_view_ratio', 0)
+    item_views = feat.get('item_views', 0)
+    cart_adds = feat.get('cart_adds', 0)
+
+    # baseline demand
+    demand = 1.0
+
+    # agent detection: high velocity → treat as high "demand" to price up
+    if velocity > 2.0:
+        demand += 10.0  # strong agent signal
+
+    # conversion intent: cart interaction → price up
+    if cart_ratio > 0.1 or cart_adds > 0:
+        demand += 5.0
+
+    # browsing depth: many views → interest signal
+    if item_views > 3:
+        demand += min(item_views, 5.0)
+
+    return min(demand, 20.0)  # cap at 20
+
+
 class StaticPricer(PricingFunction):
     """Static pricing: always return fixed base prices"""
 
@@ -25,6 +65,11 @@ class StaticPricer(PricingFunction):
             raise ValueError("Must call fit() or provide base_prices in constructor")
         return self.base_prices.copy()
 
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Static pricer uses no features, returns empty array"""
+        n = len(self.base_prices) if self.base_prices is not None else 0
+        return np.zeros((n, 0))
+
 
 class RandomPricer(PricingFunction):
     """Random pricing within bounds (for baseline comparison)"""
@@ -46,3 +91,68 @@ class RandomPricer(PricingFunction):
         if self.n_products is None:
             self.n_products = len(state_space.demand)
         return self.rng.uniform(self.price_min, self.price_max, size=self.n_products)
+
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Random pricer uses no features"""
+        n = self.n_products if self.n_products else 0
+        return np.zeros((n, 0))
+
+
+class SimpleSurgePricer(PricingFunction):
+    """
+    Rule-based surge pricer adjusting prices via demand thresholds.
+    Logic: if demand > high_threshold -> surge, if demand < low_threshold -> discount.
+    Simpler and more controllable than curve fitting approaches.
+    """
+
+    def __init__(self,
+                 base_prices: np.ndarray = None,
+                 high_threshold: int = 10,
+                 low_threshold: int = 2,
+                 surge_multiplier: float = 1.2,
+                 discount_multiplier: float = 0.9):
+        self.base_prices = base_prices
+        self.high_threshold = high_threshold
+        self.low_threshold = low_threshold
+        self.surge_multiplier = surge_multiplier
+        self.discount_multiplier = discount_multiplier
+
+    def fit(self, market_data: pd.DataFrame):
+        """Extract base prices from product catalog or historical averages"""
+        self.base_prices = market_data['base_price'].to_numpy() if 'base_price' in market_data.columns else market_data['price'].values
+        return self
+
+    def predict(self, state_space) -> np.ndarray:
+        """
+        Adjust prices based on current demand using surge rules.
+        state_space.demand: demand proxy per product (from session features)
+        state_space.prices: base prices
+        """
+        demand = np.asarray(state_space.demand) if state_space and hasattr(state_space, 'demand') else np.array([0])
+        base = np.asarray(state_space.prices) if state_space and hasattr(state_space, 'prices') else self.base_prices
+
+        if base is None:
+            base = np.ones(len(demand)) * 99.99
+
+        # ensure float dtype to allow multiplication by float multipliers
+        new_prices = base.astype(np.float64).copy()
+        high_mask = demand >= self.high_threshold
+        new_prices[high_mask] *= self.surge_multiplier
+
+        low_mask = demand <= self.low_threshold
+        new_prices[low_mask] *= self.discount_multiplier
+
+        return new_prices
+
+    def _get_features(self, state_space=None) -> np.ndarray:
+        """Extract demand and base price features for each product"""
+        if state_space is None:
+            n = len(self.base_prices) if self.base_prices is not None else 0
+            return np.zeros((n, 2))
+
+        demand = np.asarray(state_space.demand) if hasattr(state_space, 'demand') else np.array([0])
+        base = np.asarray(state_space.prices) if hasattr(state_space, 'prices') else self.base_prices
+        if base is None:
+            base = np.ones(len(demand)) * 99.99
+
+        return np.column_stack([demand, base])

experiments/procesing/providers/supabase.py

@@ -18,10 +18,17 @@ class SupabaseProvider(DataProvider):
         self.supabase: Client = create_client(self.supabase_url, self.supabase_key)
 
     def fetch_products(self, store_mode: str) -> pd.DataFrame:
-        resp = self.supabase.table(f'{store_mode}_products').select(
+        # hotel uses room_type, airline uses flight_type; select all and normalize
-            "id, room_type, date_index, metadata, availability"
+        resp = self.supabase.table(f'{store_mode}_products').select("*").execute()
-        ).execute()
+        if not resp.data:
-        return pd.DataFrame(resp.data) if resp.data else pd.DataFrame()
+            return pd.DataFrame()
+        df = pd.DataFrame(resp.data)
+        # normalize type column: hotel has room_type, airline has flight_type
+        if 'room_type' in df.columns:
+            df['product_type'] = df['room_type']
+        elif 'flight_type' in df.columns:
+            df['product_type'] = df['flight_type']
+        return df
 
     def fetch_experiments(self, experiment_ids: List[str]) -> pd.DataFrame:
         if not experiment_ids:

experiments/procesing/steps/__init__.py

@@ -1,11 +1,16 @@
 from procesing.steps.base import BaseContextStep
 from procesing.steps.fetch import FetchInteractionsStep, FetchPriceLogsStep, FetchExperimentsStep
-from procesing.steps.join import JoinExperimentsStep
+from procesing.steps.join import JoinExperimentsStep, JoinProductFeaturesStep
-from procesing.steps.augment import CreatePriceBucketsStep, AugmentEventNamesStep
+from procesing.steps.augment import CreatePriceBucketsStep, AugmentEventNamesStep, AugmentInteractionsStep
 from procesing.steps.chunk import ChunkByTimeWindowStep
 from procesing.steps.demand import ComputeDemandStep, ComputeDemandForChunksStep
-from procesing.steps.elasticity import AggregatePriceLogsStep, ComputeElasticityStep
+from procesing.steps.elasticity import AggregatePriceLogsStep
-from procesing.steps.pricing import StateSpace, BuildStateSpaceStep, FitPricingFunctionStep, PredictPricesStep
+from procesing.steps.pricing import FitPricingFunctionStep, PredictPricesStep
+from procesing.steps.session import (
+    ExtractSessionFeaturesStep, JoinLabelsStep, ValidateDataStep,
+    TemporalFeatureStep, BehavioralFeatureStep, ProductFeatureStep, UserAgentFeatureStep,
+    _extract_features_for_session
+)
 
 __all__ = [
     'BaseContextStep',
@@ -13,15 +18,22 @@ __all__ = [
     'FetchPriceLogsStep',
     'FetchExperimentsStep',
     'JoinExperimentsStep',
+    'JoinProductFeaturesStep',
     'CreatePriceBucketsStep',
     'AugmentEventNamesStep',
+    'AugmentInteractionsStep',
     'ChunkByTimeWindowStep',
     'ComputeDemandStep',
     'ComputeDemandForChunksStep',
     'AggregatePriceLogsStep',
-    'ComputeElasticityStep',
-    'StateSpace',
-    'BuildStateSpaceStep',
     'FitPricingFunctionStep',
     'PredictPricesStep',
+    'ExtractSessionFeaturesStep',
+    'JoinLabelsStep',
+    'ValidateDataStep',
+    'TemporalFeatureStep',
+    'BehavioralFeatureStep',
+    'ProductFeatureStep',
+    'UserAgentFeatureStep',
+    '_extract_features_for_session',
 ]

experiments/procesing/steps/augment.py

@@ -2,6 +2,93 @@ import numpy as np
 import pandas as pd
 from procesing.steps.base import BaseContextStep
 
+
+class AugmentInteractionsStep(BaseContextStep):
+    """
+    Consolidated step: create price buckets, augment event names, join experiments.
+    Input: (interactions_df, price_logs_df)
+    Output: enriched interactions_df
+    """
+
+    def transform(self, data: tuple):
+        interactions_df, price_logs_df = data
+
+        if interactions_df.empty:
+            return interactions_df
+
+        # Step 1: Create price buckets
+        interactions_df = self._create_price_buckets(interactions_df)
+
+        # Step 2: Augment event names
+        interactions_df = self._augment_event_names(interactions_df)
+
+        # Step 3: Join experiments (optional)
+        if 'experimentId' in interactions_df.columns:
+            interactions_df = self._join_experiments(interactions_df)
+
+        return interactions_df
+
+    def _create_price_buckets(self, df: pd.DataFrame):
+        """Create price bucket labels from price data"""
+        if 'metadata_price' not in df.columns:
+            df['price_bucket'] = ""
+            return df
+
+        n_buckets = self.context.config.get('n_price_buckets', 5)
+
+        if df['metadata_price'].notnull().sum() > 0:
+            try:
+                price_buckets = pd.qcut(
+                    df['metadata_price'],
+                    q=n_buckets,
+                    labels=[f"PB_{i+1}" for i in range(n_buckets)],
+                    duplicates='drop'
+                )
+            except ValueError:
+                # fallback for insufficient unique values
+                price_buckets = df['metadata_price'].apply(
+                    lambda x: f"P_{int(x)}" if pd.notnull(x) else ""
+                )
+        else:
+            price_buckets = pd.Series([""] * len(df), index=df.index)
+
+        df['price_bucket'] = price_buckets
+        return df
+
+    def _augment_event_names(self, df: pd.DataFrame):
+        """Augment event names with product and price bucket schema"""
+        # Create schema: _productId@price_bucket
+        has_product = df.get('productId', pd.Series()).notnull()
+        has_bucket = df.get('price_bucket', pd.Series()).notnull()
+
+        df['metadata_schema'] = np.where(
+            has_product & has_bucket,
+            "_" + df['productId'].astype(str) + "@" + df['price_bucket'].astype(str),
+            ""
+        )
+
+        df['eventName'] = df['eventName'] + df['metadata_schema']
+        return df
+
+    def _join_experiments(self, df: pd.DataFrame):
+        """Join experiment metadata if experimentId present"""
+        exp_ids = df['experimentId'].dropna().unique().tolist()
+        if not exp_ids:
+            return df
+
+        experiments_df = self.context.provider.fetch_experiments(exp_ids)
+        if experiments_df.empty:
+            return df
+
+        return df.merge(
+            experiments_df,
+            left_on='experimentId',
+            right_on='id',
+            how='left',
+            suffixes=('', '_exp')
+        )
+
+
 class CreatePriceBucketsStep(BaseContextStep):
     """Create price bucket labels from price data"""
 

experiments/procesing/steps/base.py

@@ -1,6 +1,7 @@
 from abc import ABC, abstractmethod
 from sklearn.base import BaseEstimator, TransformerMixin
 from procesing.context import PipelineContext
+from typing import Any
 
 class BaseContextStep(BaseEstimator, TransformerMixin, ABC):
     """
@@ -16,7 +17,7 @@ class BaseContextStep(BaseEstimator, TransformerMixin, ABC):
         return self
 
     @abstractmethod
-    def transform(self, X):
+    def transform(self, X) -> Any:
         """Transform input using context. Must be implemented by subclass."""
         pass
 

experiments/procesing/steps/elasticity.py

@@ -7,16 +7,16 @@ class AggregatePriceLogsStep(BaseContextStep):
     """
     Aggregate price logs into time windows using VECTORIZED operations.
     Input: price_logs_df
-    Output: list of price chunks with [productId, price]
+    Output: DataFrame with columns [productId, price]
     """
 
     def transform(self, price_logs_df: pd.DataFrame):
         if price_logs_df.empty:
-            return []
+            return pd.DataFrame(columns=['productId', 'price'])
 
         df = price_logs_df.copy()
         ts_col = self.context.config.get('ts_col', 'ts')
-        window_size = self.context.window_size
+        #window_size = self.context.window_size WE ARE NOT USING CHUNKS ANYMORE
 
         # ensure datetime
         if not pd.api.types.is_datetime64_any_dtype(df[ts_col]):
@@ -24,230 +24,19 @@ class AggregatePriceLogsStep(BaseContextStep):
 
         df = df.sort_values([ts_col, 'productId'])
         products = self.context.products
-        unique_products = products['id'].unique()
+        # get base price from metadata if available 1) read the metadata col as json and get the base_price
-
+        products['base_price'] = products.apply(
-        # VECTORIZED: group by product, resample by time window, compute mean
+            lambda row: row['metadata'].get('base_price', 0) if isinstance(row['metadata'], dict) else 0,
-        df_indexed = df.set_index(ts_col)
+            axis=1
-
-        windowed = (
-            df_indexed
-            .groupby('productId')['price']
-            .resample(window_size)
-            .mean()
-            .reset_index()
         )
 
-        # forward fill missing windows (carry last known price)
+        unique_products = products['id'].unique()
-        windowed = windowed.sort_values([ts_col, 'productId'])
-        windowed['price'] = windowed.groupby('productId')['price'].ffill()
-        windowed = windowed.dropna(subset=['price'])
 
-        # group into chunks by window
+        df_indexed = df.set_index(ts_col)
-        chunks = []
+        # we return a df of average price per product over the entire period
-        for window_start, group in windowed.groupby(ts_col):
+        # TODO: maybe consider different opration to handle price aggregation over time
-            price_vector = group[['productId', 'price']].copy()
+        avg_prices = df_indexed.groupby('productId')['price'].mean().reindex(unique_products, fill_value=0).reset_index()
-
+        avg_prices.columns = ['productId', 'price']
-            # fill missing products with last known price before this window
+        # fill 0s with base_price from products
-            missing_products = set(unique_products) - set(price_vector['productId'])
+        base_price_map = products.set_index('id')['base_price'].to_dict()
-            if missing_products:
+        return avg_prices
-                for pid in missing_products:
-                    last_price = df_indexed[
-                        (df_indexed['productId'] == pid) &
-                        (df_indexed.index < window_start)
-                    ]['price']
-
-                    if not last_price.empty:
-                        price_vector = pd.concat([
-                            price_vector,
-                            pd.DataFrame({'productId': [pid], 'price': [last_price.iloc[-1]]})
-                        ], ignore_index=True)
-
-            if not price_vector.empty:
-                chunks.append({
-                    'window_start': window_start,
-                    'window_end': window_start + pd.Timedelta(window_size),
-                    'price_vector': price_vector
-                })
-
-        return chunks
-
-
-class ComputeElasticityStep(BaseContextStep):
-    """
-    Compute price elasticity from demand and price chunks.
-    Input: (demand_chunks, price_chunks)
-    Output: elasticity_df [productId, elasticity, std_error, n_obs]
-    """
-
-    def transform(self, chunk_tuple: tuple):
-        demand_chunks, price_chunks = chunk_tuple
-
-        method = self.context.config.get('elasticity_method', 'point')
-        min_obs = self.context.config.get('min_observations', 2)
-
-        products = self.context.products
-        all_product_ids = products['id'].unique()
-
-        # align chunks by window_start
-        aligned = self._align_chunks(demand_chunks, price_chunks)
-
-        if not aligned:
-            return pd.DataFrame({
-                'productId': all_product_ids,
-                'elasticity': 0.0,
-                'std_error': 0.0,
-                'n_obs': 0
-            })
-
-        # build time series per product
-        product_series = self._build_timeseries(aligned)
-
-        # compute elasticity per product
-        elasticities = []
-        for pid, series in product_series.items():
-            if len(series) < min_obs:
-                elasticities.append({
-                    'productId': pid,
-                    'elasticity': 0.0,
-                    'std_error': 0.0,
-                    'n_obs': len(series)
-                })
-                continue
-
-            elast = self._compute_elasticity(series, method)
-            elasticities.append({
-                'productId': pid,
-                'elasticity': elast['value'],
-                'std_error': elast.get('std_error', 0.0),
-                'n_obs': len(series)
-            })
-
-        result_df = pd.DataFrame(elasticities)
-
-        # fill missing products with zero elasticity
-        observed_pids = set(result_df['productId'])
-        missing_pids = [p for p in all_product_ids if p not in observed_pids]
-
-        if missing_pids:
-            missing_df = pd.DataFrame({
-                'productId': missing_pids,
-                'elasticity': 0.0,
-                'std_error': 0.0,
-                'n_obs': 0
-            })
-            result_df = pd.concat([result_df, missing_df], ignore_index=True)
-
-        return result_df
-
-    def _align_chunks(self, demand_chunks: List[Dict], price_chunks: List[Dict]):
-        """Align demand and price chunks by window_start"""
-        price_lookup = {c['window_start']: c for c in price_chunks}
-        aligned = []
-
-        for dc in demand_chunks:
-            ws = dc['window_start']
-            if ws in price_lookup:
-                aligned.append({
-                    'window_start': ws,
-                    'window_end': dc['window_end'],
-                    'demand': dc['demand_vector'],
-                    'prices': price_lookup[ws]['price_vector']
-                })
-
-        return aligned
-
-    def _build_timeseries(self, aligned: List[Dict]):
-        """Build time series [timestamp, price, quantity] per product"""
-        series_by_product = {}
-
-        for chunk in aligned:
-            merged = chunk['demand'].merge(chunk['prices'], on='productId', how='inner')
-
-            for _, row in merged.iterrows():
-                pid = row['productId']
-                if pid not in series_by_product:
-                    series_by_product[pid] = []
-
-                series_by_product[pid].append({
-                    'timestamp': chunk['window_start'],
-                    'price': row['price'],
-                    'quantity': row['demand_score']
-                })
-
-        return series_by_product
-
-    def _compute_elasticity(self, series: List[Dict], method: str):
-        """Compute point or arc elasticity"""
-        prices = np.array([s['price'] for s in series])
-        quantities = np.array([s['quantity'] for s in series])
-
-        # filter out zero/negative values
-        valid = (prices > 0) & (quantities > 0)
-        if valid.sum() < 2:
-            return {'value': 0.0, 'std_error': 0.0}
-
-        prices = prices[valid]
-        quantities = quantities[valid]
-
-        if method == 'point':
-            return self._point_elasticity(prices, quantities)
-        elif method == 'arc':
-            return self._arc_elasticity(prices, quantities)
-        else:
-            raise ValueError(f"Unknown elasticity method: {method}")
-
-    def _point_elasticity(self, prices: np.ndarray, quantities: np.ndarray):
-        """Point elasticity via log-log regression: log(Q) = a + b*log(P), elasticity = b"""
-        if len(prices) < 2:
-            return {'value': 0.0, 'std_error': 0.0}
-
-        log_p = np.log(prices)
-        log_q = np.log(quantities)
-
-        if log_p.std() == 0:
-            return {'value': 0.0, 'std_error': 0.0}
-
-        cov = np.cov(log_p, log_q)[0, 1]
-        var = np.var(log_p)
-        b = cov / var
-
-        # std error estimate
-        if len(prices) > 2:
-            residuals = log_q - (log_q.mean() + b * (log_p - log_p.mean()))
-            mse = (residuals ** 2).sum() / (len(prices) - 2)
-            se_b = np.sqrt(mse / (len(prices) * var))
-        else:
-            se_b = 0.0
-
-        return {'value': b, 'std_error': se_b}
-
-    def _arc_elasticity(self, prices: np.ndarray, quantities: np.ndarray):
-        """Arc elasticity: average period-over-period elasticity"""
-        elasticities = []
-
-        for i in range(1, len(prices)):
-            p1, p2 = prices[i-1], prices[i]
-            q1, q2 = quantities[i-1], quantities[i]
-
-            p_avg = (p1 + p2) / 2
-            q_avg = (q1 + q2) / 2
-
-            if p_avg == 0 or q_avg == 0:
-                continue
-
-            delta_p = p2 - p1
-            delta_q = q2 - q1
-
-            if delta_p == 0:
-                continue
-
-            e = (delta_q / q_avg) / (delta_p / p_avg)
-            elasticities.append(e)
-
-        if not elasticities:
-            return {'value': 0.0, 'std_error': 0.0}
-
-        return {
-            'value': np.mean(elasticities),
-            'std_error': np.std(elasticities) / np.sqrt(len(elasticities))
-        }

experiments/procesing/steps/fetch.py

@@ -2,7 +2,11 @@ import pandas as pd
 from procesing.steps.base import BaseContextStep
 
 class FetchInteractionsStep(BaseContextStep):
-    """Fetch raw interaction data from Kafka topic"""
+    """Fetch raw interaction data from Kafka topic with optional time and store_mode filtering"""
+
+    def __init__(self, context, lookback: str = None):
+        super().__init__(context)
+        self.lookback = lookback
 
     def transform(self, X=None):
         df = self.context.provider.fetch_kafka_topic('user-interactions')
@@ -17,19 +21,50 @@ class FetchInteractionsStep(BaseContextStep):
             )
 
         df = df.dropna(subset=['eventName'])
+        # drop all where page has /admin/
+        df = df[~df['page'].str.contains('/admin/', na=False)]
+
+        # filter by store_mode from context
+        if 'storeMode' in df.columns:
+            df = df[df['storeMode'] == self.context.store_mode]
 
         # Remap dateIndex if present
         if 'metadata_dateIndex' in df.columns:
             df['dateIndex'] = df['metadata_dateIndex'].astype('Int64')
 
+        # Apply time filtering if lookback specified
+        if self.lookback and 'ts' in df.columns:
+            df['ts'] = pd.to_datetime(df['ts'])
+            cutoff = pd.Timestamp.now() - pd.Timedelta(self.lookback)
+            df = df[df['ts'] >= cutoff]
+
         return df
 
 
 class FetchPriceLogsStep(BaseContextStep):
-    """Fetch price log data from Kafka topic"""
+    """Fetch price log data from Kafka topic with optional time and store_mode filtering"""
+
+    def __init__(self, context, lookback: str = None):
+        super().__init__(context)
+        self.lookback = lookback
 
     def transform(self, X=None):
-        return self.context.provider.fetch_kafka_topic('price-logs')
+        df = self.context.provider.fetch_kafka_topic('price-logs')
+
+        if df.empty:
+            return df
+
+        # filter by store_mode from context
+        if 'storeMode' in df.columns:
+            df = df[df['storeMode'] == self.context.store_mode]
+
+        # Apply time filtering if lookback specified
+        if self.lookback and 'ts' in df.columns:
+            df['ts'] = pd.to_datetime(df['ts'])
+            cutoff = pd.Timestamp.now() - pd.Timedelta(self.lookback)
+            df = df[df['ts'] >= cutoff]
+
+        return df
 
 
 class FetchExperimentsStep(BaseContextStep):

experiments/procesing/steps/join.py

@@ -32,3 +32,27 @@ class JoinExperimentsStep(BaseContextStep):
         })
 
         return interactions_df.merge(experiments_df, on='experimentId', how='left')
+
+class JoinProductFeaturesStep(BaseContextStep):
+    """Join product features to interactions"""
+
+    def transform(self, data: tuple):
+        """
+        Args:
+            data: (interactions_df, products_df)
+        Returns:
+            merged interactions dataframe
+        """
+        demand_df, price_df = data
+
+        # get base prices from products if available
+        products = self.context.products
+        products['base_price'] = products.apply(
+            lambda row: float(row['metadata'].get('base_price', 0.0)) if isinstance(row['metadata'], dict) else 0,
+            axis=1
+        )
+        products = products[['id', 'base_price']].rename(columns={'id': 'productId'})
+
+        if price_df.empty:
+            return demand_df
+        return demand_df.merge(price_df, on='productId', how='left').merge(products, on='productId', how='left')

experiments/procesing/steps/pricing.py

@@ -2,128 +2,34 @@ import numpy as np
 import pandas as pd
 from typing import Optional, List, Dict, Any
 from dataclasses import dataclass, field
+from procesing.pricers.simple import StaticPricer
 from procesing.steps.base import BaseContextStep
 from procesing.pricers import ElasticityBasedPricer
 
-@dataclass
+class State:
-class StateSpace:
+    def __init__(self,
-    """
+                 last_action : str,
-    State representation for pricing functions.
+                 last_productId : str,
+                 last_price : float,
+                 session_features : np.ndarray
+                 ):
+        pass
 
-    Components:
-        Q_t: demand ∈ R^n (current demand signal per product)
-        P_t: prices ∈ R^n (current/base prices)
-        S_t: session_features (behavioral signals, interaction data)
-        H_t: history = {Q_{t-k}, P_{t-k}, S_{t-k}} for k in [1, history_length]
-
-    Additionally stores:
-        - product_ids: product identifiers (n,)
-        - elasticity: price elasticity per product (n,)
-        - metadata: arbitrary context (experiment_id, timestamp, etc.)
-    """
-    demand: np.ndarray  # Q_t ∈ R^n
-    prices: np.ndarray  # P_t ∈ R^n
-    session_features: pd.DataFrame = field(default_factory=pd.DataFrame)  # S_t
-
-    # augmented state components
-    product_ids: Optional[np.ndarray] = None
-    elasticity: Optional[np.ndarray] = None
-
-    # historical trajectory H_t = {(Q_{t-k}, P_{t-k}, S_{t-k})}
-    history: List[Dict[str, Any]] = field(default_factory=list)
-
-    # metadata for context
-    metadata: Dict[str, Any] = field(default_factory=dict)
-
-    def __post_init__(self):
-        """Validate dimensions."""
-        n = len(self.demand)
-        assert len(self.prices) == n, "demand and prices must have same dimension"
-        if self.elasticity is not None:
-            assert len(self.elasticity) == n, "elasticity must match dimension"
-        if self.product_ids is not None:
-            assert len(self.product_ids) == n, "product_ids must match dimension"
-
-    @property
-    def n_products(self) -> int:
-        """Number of products in state space."""
-        return len(self.demand)
-
-    def add_history(self, q: np.ndarray, p: np.ndarray, s: pd.DataFrame, max_length: int = 10):
-        """Append historical state to trajectory H_t."""
-        self.history.append({'demand': q, 'prices': p, 'session_features': s})
-        if len(self.history) > max_length:
-            self.history.pop(0)
-
-    def get_history_window(self, k: int = 5) -> List[Dict[str, Any]]:
-        """Retrieve last k historical states."""
-        return self.history[-k:] if len(self.history) >= k else self.history
-
-
-class BuildStateSpaceStep(BaseContextStep):
-    """
-    Build state space from elasticity, demand, and price data.
-
-    Input: elasticity_df [productId, elasticity, ...], optional demand_df
-    Output: StateSpace instance with Q_t, P_t, elasticity, product_ids
-    """
-
-    def transform(self, elasticity_df: pd.DataFrame, demand_df: Optional[pd.DataFrame] = None):
-        products = self.context.products
-
-        # extract base prices from product metadata
-        products_with_prices = products.copy()
-        if 'metadata' in products_with_prices.columns:
-            products_with_prices['base_price'] = products_with_prices['metadata'].apply(
-                lambda m: m.get('base_price', 0) if isinstance(m, dict) else 0
-            )
-        else:
-            products_with_prices['base_price'] = 0
-
-        # merge with elasticity
-        merged = products_with_prices[['id', 'base_price']].rename(
-            columns={'id': 'productId'}
-        ).merge(
-            elasticity_df[['productId', 'elasticity']],
-            on='productId',
-            how='left'
-        ).fillna({'elasticity': 0.0, 'base_price': 0.0})
-
-        # merge with demand if provided, else use default
-        if demand_df is not None and 'demand' in demand_df.columns:
-            merged = merged.merge(
-                demand_df[['productId', 'demand']],
-                on='productId',
-                how='left'
-            ).fillna({'demand': 0.0})
-            demand_vector = merged['demand'].values
-        else:
-            # default: uniform demand or use elasticity as proxy
-            demand_vector = np.ones(len(merged)) * 10.0
-
-        return StateSpace(
-            demand=demand_vector,
-            prices=merged['base_price'].values,
-            session_features=pd.DataFrame(),
-            product_ids=merged['productId'].values,
-            elasticity=merged['elasticity'].values,
-            metadata={'timestamp': pd.Timestamp.now().isoformat()}
-        )
 
 
 class FitPricingFunctionStep(BaseContextStep):
     """
-    Fit pricing function using elasticity data.
+    Fit pricing function using data.
-    Input: elasticity_df
+    Input: pricing_data
     Output: fitted pricing function instance
     """
 
-    def transform(self, elasticity_df: pd.DataFrame):
+    def transform(self, pricing_data: pd.DataFrame):
-        pricing_class = self.context.config.get('pricing_function_class', ElasticityBasedPricer)
+        pricing_class = self.context.config.get('pricing_function_class', StaticPricer)
         pricing_params = self.context.config.get('pricing_function_params', {})
 
         pricer = pricing_class(**pricing_params)
-        pricer.fit(elasticity_df)
+        pricer.fit(pricing_data)
 
         return pricer
 

experiments/procesing/steps/session.py

@@ -1,114 +1,262 @@
 """
-Session feature extraction for S_t component of state space.
+Session feature extraction for ML training pipeline.
-Computes behavioral signals from interaction data already in pipeline.
 """
 import pandas as pd
 import numpy as np
-from typing import Optional, Dict, Any
+import re
-from collections import Counter
+from typing import Dict, Any
 from procesing.steps.base import BaseContextStep
 
+EVENT_CATS = {
+    'page_view': ['page_view'],
+    'item_view': ['view_item_page', 'learn_more_about_item'],
+    'cart_add': ['add_item_to_cart'],
+    'purchase': ['purchase', 'checkout_complete'],
+    'hover': ['hover_over_title', 'hover_over_paragraph', 'hover_over_link', 'hover_over_button'],
+    # 'filter': ['filter', 'search', 'apply_filter'],
+}
+HEADLESS_RE = re.compile(r'HeadlessChrome|Headless|PhantomJS', re.I)
+AUTOMATION_RE = re.compile(r'Selenium|Playwright|Puppeteer|WebDriver|chromedriver|geckodriver', re.I)
+BROWSER_PATTERNS = [('Chrome', r'Chrome/[\d.]+'), ('Firefox', r'Firefox/[\d.]+'),
+                    ('Safari', r'Safari/[\d.]+'), ('Edge', r'Edg/[\d.]+')]
+
+
+def _get_browser(s: str) -> str:
+    if pd.isna(s): return 'Unknown'
+    for name, pat in BROWSER_PATTERNS:
+        if re.search(pat, s): return name
+    return 'Other'
+
+
+class TemporalFeatureStep(BaseContextStep):
+    """Vectorized time-based features: durations, velocities, gaps."""
+
+    def __init__(self, context, timeout_sec: float = 900, velocity_window: str = '5min'):
+        super().__init__(context)
+        self.timeout_sec = timeout_sec
+        self.velocity_window = velocity_window
+
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame:
+        df = X.copy()
+        if df.empty or 'ts' not in df.columns:
+            return pd.DataFrame(columns=pd.Series(['sessionId']))
+
+        df['ts_dt'] = pd.to_datetime(df['ts'])
+        df = df.sort_values(['sessionId', 'ts_dt'])
+        df['time_diff'] = df.groupby('sessionId')['ts_dt'].diff().dt.total_seconds()
+        df['active_diff'] = df['time_diff'].where(df['time_diff'] <= self.timeout_sec, 0)
+
+        agg = df.groupby('sessionId').agg(
+            session_duration_sec=('active_diff', 'sum'),
+            total_interactions=('sessionId', 'count'),
+            avg_time_between_events=('time_diff', 'mean'),
+            std_time_between_events=('time_diff', 'std'),
+            min_time_between_events=('time_diff', 'min'),
+            session_start_hour=('ts_dt', lambda x: x.min().hour),
+        ).reset_index()
+        agg['std_time_between_events'] = agg['std_time_between_events'].fillna(0)
+        agg['interaction_velocity'] = np.where(
+            agg['session_duration_sec'] > 0,
+            (agg['total_interactions'] / agg['session_duration_sec']) * 60, 0)
+
+        vel = df.set_index('ts_dt').groupby('sessionId').resample(self.velocity_window, include_groups=False).size()
+        max_velocity = vel.groupby('sessionId').max().rename('max_velocity_5min')
+        agg = agg.merge(max_velocity, on='sessionId', how='left')
+        agg['max_velocity_5min'] = agg['max_velocity_5min'].fillna(0)
+        return agg
+
+
+class BehavioralFeatureStep(BaseContextStep):
+    """Vectorized event counts and ratios per session."""
+
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame:
+        df = X.copy()
+        if df.empty or 'eventName' not in df.columns:
+            return pd.DataFrame(columns=pd.Series(['sessionId']))
+
+        for cat, events in EVENT_CATS.items():
+            df[f'is_{cat}'] = df['eventName'].isin(events)
+        df['is_hover'] = df['is_hover'] | df['eventName'].str.startswith('hover_over_')
+
+        agg = df.groupby('sessionId').agg(
+            total_events=('eventName', 'count'), unique_pages=('page', 'nunique'),
+            page_views=('is_page_view', 'sum'), item_views=('is_item_view', 'sum'),
+            cart_adds=('is_cart_add', 'sum'), purchases=('is_purchase', 'sum'),
+            hover_events=('is_hover', 'sum'),
+            # filter_events=('is_filter', 'sum'),
+        ).reset_index()
+        agg['cart_to_view_ratio'] = np.where(agg['item_views'] > 0, agg['cart_adds'] / agg['item_views'], 0)
+        agg['conversion_rate'] = np.where(agg['item_views'] > 0, agg['purchases'] / agg['item_views'], 0)
+        agg['hover_intensity'] = np.where(agg['total_events'] > 0, agg['hover_events'] / agg['total_events'], 0)
+        return agg
+
+
+class ProductFeatureStep(BaseContextStep):
+    """Vectorized product interaction features: diversity, depth, price sensitivity."""
+
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame:
+        df = X.copy()
+        if df.empty:
+            return pd.DataFrame(columns=pd.Series(['sessionId']))
+        price_col = next((c for c in ['metadata_base_price', 'metadata_price', 'base_price'] if c in df.columns), None)
+        df['price_seen'] = pd.to_numeric(df[price_col], errors='coerce') if price_col else np.nan
+
+        prod_df = df[df['productId'].notna()]
+        if prod_df.empty:
+            return pd.DataFrame(columns=pd.Series(['sessionId', 'unique_products_viewed', 'product_view_depth', 'avg_price_seen', 'min_price_seen', 'max_price_seen', 'price_range']))
+
+        agg = prod_df.groupby('sessionId').agg(
+            unique_products_viewed=('productId', 'nunique'),
+            product_view_depth=('productId', lambda x: x.value_counts().iloc[0] if len(x) > 0 else 0),
+            avg_price_seen=('price_seen', 'mean'), min_price_seen=('price_seen', 'min'),
+            max_price_seen=('price_seen', 'max'),
+        ).reset_index()
+        agg['price_range'] = (agg['max_price_seen'] - agg['min_price_seen']).fillna(0)
+        return agg
+
+
+class UserAgentFeatureStep(BaseContextStep):
+    """Parse userAgent into bot-detection signals."""
+
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame|pd.Series:
+        df = X.copy()
+        if df.empty or 'userAgent' not in df.columns:
+            return pd.DataFrame(columns=pd.Series(['sessionId']))
+
+        ua = df.groupby('sessionId')['userAgent'].first().reset_index()
+        ua['is_headless'] = ua['userAgent'].str.contains(HEADLESS_RE, na=False)
+        ua['is_automation'] = ua['userAgent'].str.contains(AUTOMATION_RE, na=False)
+        ua['browser_family'] = ua['userAgent'].apply(_get_browser)
+        return ua[['sessionId', 'is_headless', 'is_automation', 'browser_family']]
+
 
 class ExtractSessionFeaturesStep(BaseContextStep):
     """
-    Extract session-level behavioral features from interaction logs.
+    Vectorized session feature extraction - replaces O(n^2) per-row loop.
-
+    Input: interactions_df
-    Input: interactions_df (user-interactions from earlier pipeline step)
+    Output: session-level feature matrix
-    Output: session_features DataFrame [sessionId, feature_1, feature_2, ...]
+    THIS is our main mapping from tau (trajectory) to some features vector theta - we need to do this very well. This is what will go into demand esimation.
-
-    Features computed:
-        - total_interactions: count of all events
-        - page_views, item_views, searches, cart_adds: event type counts
-        - hovers: hover event counts
-        - unique_products_viewed: distinct product IDs
-        - interaction_velocity: events per minute
-        - session_duration_sec: time span of session
-        - avg_time_between_events: mean inter-event time
-        - product_view_depth: max views for single product (attention signal)
     """
 
-    def transform(self, interactions_df: pd.DataFrame) -> pd.DataFrame:
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame:
-        if interactions_df.empty:
+        if X.empty:
             return pd.DataFrame()
+        df = X.copy()
 
-        # ensure timestamp column
+        # run all feature steps and merge on sessionId
-        if 'ts' in interactions_df.columns:
+        temporal = TemporalFeatureStep(self.context).transform(df)
-            interactions_df = interactions_df.copy()
+        behavioral = BehavioralFeatureStep(self.context).transform(df)
-            interactions_df['ts'] = pd.to_datetime(interactions_df['ts'])
+        product = ProductFeatureStep(self.context).transform(df)
+        ua = UserAgentFeatureStep(self.context).transform(df)
 
-        # group by session and compute features
+        result = temporal
-        session_features = []
+        for other in [behavioral, product, ua]:
-        for session_id, session_df in interactions_df.groupby('sessionId'):
+            if not other.empty and 'sessionId' in other.columns:
-            features = self._extract_features_for_session(session_id, session_df)
+                result = result.merge(other, on='sessionId', how='left')
-            session_features.append(features)
 
-        return pd.DataFrame(session_features)
+        # carry forward experimentId for label joining
+        if 'experimentId' in df.columns:
+            exp_map = df.groupby('sessionId')['experimentId'].first()
+            result = result.merge(exp_map, on='sessionId', how='left')
 
-    def _extract_features_for_session(self, session_id: str, session_df: pd.DataFrame) -> Dict[str, Any]:
+        return result
-        """Compute features for single session."""
-        features = {'sessionId': session_id}
-
-        # basic counts
-        features['total_interactions'] = len(session_df)
-
-        event_counts = session_df['eventName'].value_counts().to_dict()
-        features['page_views'] = event_counts.get('page_view', 0) + event_counts.get('view_item_page', 0)
-        features['item_views'] = event_counts.get('view_item_page', 0)
-        features['searches'] = event_counts.get('search', 0)
-        features['cart_adds'] = event_counts.get('add_item_to_cart', 0)
-
-        # hover events
-        hover_events = ['hover_over_title', 'hover_over_paragraph', 'hover_over_link', 'hover_over_button']
-        features['hovers'] = sum(event_counts.get(ev, 0) for ev in hover_events)
-
-        # product-level signals
-        product_ids = session_df['productId'].dropna()
-        features['unique_products_viewed'] = product_ids.nunique()
-
-        if len(product_ids) > 0:
-            product_view_counts = Counter(product_ids)
-            features['product_view_depth'] = max(product_view_counts.values())
-        else:
-            features['product_view_depth'] = 0
-
-        # temporal features
-        if 'ts' in session_df.columns:
-            timestamps = session_df['ts'].sort_values()
-            features['session_duration_sec'] = (timestamps.max() - timestamps.min()).total_seconds()
-
-            if features['session_duration_sec'] > 0:
-                features['interaction_velocity'] = (features['total_interactions'] / features['session_duration_sec']) * 60
-            else:
-                features['interaction_velocity'] = 0.0
-
-            # inter-event timing
-            if len(timestamps) > 1:
-                time_diffs = timestamps.diff().dropna().dt.total_seconds()
-                features['avg_time_between_events'] = time_diffs.mean()
-                features['std_time_between_events'] = time_diffs.std()
-            else:
-                features['avg_time_between_events'] = 0.0
-                features['std_time_between_events'] = 0.0
-        else:
-            features['session_duration_sec'] = 0.0
-            features['interaction_velocity'] = 0.0
-            features['avg_time_between_events'] = 0.0
-            features['std_time_between_events'] = 0.0
-
-        # cart/conversion signals
-        features['cart_to_view_ratio'] = features['cart_adds'] / features['item_views'] if features['item_views'] > 0 else 0.0
-
-        return features
 
 
-class FilterSessionInteractionsStep(BaseContextStep):
+class JoinLabelsStep(BaseContextStep):
     """
-    Filter interactions DataFrame to specific session.
+    Join experiment labels to session features.
-
+    Input: (features_df, experiments_df) or features_df (fetches experiments)
-    Input: (interactions_df, session_id)
+    Output: labeled feature matrix with is_agent column
-    Output: interactions_df filtered to session_id
     """
 
-    def transform(self, data: tuple) -> pd.DataFrame:
+    def transform(self, X : tuple) -> pd.DataFrame:
-        interactions_df, session_id = data
+        data = X;
-        return interactions_df[interactions_df['sessionId'] == session_id].copy()
+        if isinstance(data, tuple):
+            features_df, experiments_df = data
+        else:
+            features_df = data
+            if 'experimentId' not in features_df.columns:
+                return features_df
+            exp_ids = features_df['experimentId'].dropna().unique().tolist()
+            experiments_df = self.context.provider.fetch_experiments(exp_ids) if exp_ids else pd.DataFrame()
+
+        if features_df.empty:
+            return features_df
+        if experiments_df.empty:
+            features_df['is_agent'] = np.nan
+            return features_df
+
+        exp = experiments_df.copy()
+        if 'id' in exp.columns:
+            exp = exp.rename(columns={'id': 'experimentId'})
+        if 'xp_human_only' in exp.columns:
+            exp['is_agent'] = ~exp['xp_human_only']
+
+        cols = ['experimentId'] + [c for c in ['is_agent', 'xp_human_only', 'xp_market_mode'] if c in exp.columns]
+        return features_df.merge(exp[cols].drop_duplicates(), on='experimentId', how='left')
+
+
+class ValidateDataStep(BaseContextStep):
+    """
+    Data quality checks before training.
+    Input: df
+    Output: df (unchanged, but logs validation report to context)
+    """
+    REQUIRED = ['sessionId', 'eventName', 'ts']
+
+    def transform(self, X: pd.DataFrame) -> pd.DataFrame:
+        df = X.copy()
+        report = {'status': 'valid', 'rows': len(df), 'sessions': 0}
+        if df.empty:
+            report['status'] = 'empty'
+            self.context.cache('validation_report', report)
+            return df
+
+        missing = [c for c in self.REQUIRED if c not in df.columns]
+        if missing:
+            report['status'] = 'invalid'
+            report['missing_cols'] = missing
+
+        report['sessions'] = df['sessionId'].nunique() if 'sessionId' in df.columns else 0
+        report['null_sessions'] = int(df['sessionId'].isna().sum()) if 'sessionId' in df.columns else 0
+        if 'experimentId' in df.columns:
+            report['null_experiments'] = int(df['experimentId'].isna().sum())
+
+        self.context.cache('validation_report', report)
+        return df
+
+
+# legacy compat - kept for backwards compatibility with existing code
+def _extract_features_for_session(session_df: pd.DataFrame, session_timeout_sec: float = 900) -> Dict[str, Any]:
+    """Single-session feature extraction (legacy interface)."""
+    defaults = {k: 0 for k in ['total_interactions', 'page_views', 'item_views', 'searches',
+                               'cart_adds', 'hovers', 'unique_products_viewed', 'product_view_depth',
+                               'session_duration_sec', 'interaction_velocity',
+                               'avg_time_between_events', 'std_time_between_events', 'cart_to_view_ratio']}
+    if session_df.empty:
+        return defaults
+
+    session_df = session_df.copy()
+    if 'sessionId' not in session_df.columns:
+        session_df['sessionId'] = 'tmp'
+
+    # use a dummy context for the steps
+    class DummyCtx: config = {} # should maybe inherit but whatever
+    ctx = DummyCtx()
+
+    t = TemporalFeatureStep(ctx, timeout_sec=session_timeout_sec).transform(session_df)
+    b = BehavioralFeatureStep(ctx).transform(session_df)
+    p = ProductFeatureStep(ctx).transform(session_df)
+
+    result = {}
+    for df in [t, b, p]:
+        if not df.empty:
+            for col in df.columns:
+                if col != 'sessionId':
+                    result[col] = df[col].iloc[0] if len(df) > 0 else 0
+
+    remap = {'hover_events': 'hovers', 'filter_events': 'searches', 'unique_pages': 'unique_pages_visited'}
+    for old, new in remap.items():
+        if old in result:
+            result[new] = result.pop(old)
+    return result

experiments/procesing/tests/conftest.py

@@ -144,7 +144,7 @@ def mock_price_logs_raw_kafka():
                     'price': 162.47,
                     'sessionId': 'd423ce8a-77aa-4c9a-94d4-d1adddcc3472',
                     'experimentId': '53aefd07-f66a-4d7f-ba8b-7ea1fc562d35',
-                    'storeMode': 'shop',
+                    'storeMode': 'hotel',
                     'ts': '2025-11-25T21:05:57.967Z'
                 }
             }
@@ -157,7 +157,7 @@ def mock_price_logs_raw_kafka():
                     'price': 743.49,
                     'sessionId': 'd423ce8a-77aa-4c9a-94d4-d1adddcc3472',
                     'experimentId': '53aefd07-f66a-4d7f-ba8b-7ea1fc562d35',
-                    'storeMode': 'shop',
+                    'storeMode': 'hotel',
                     'ts': '2025-11-25T21:05:57.993Z'
                 }
             }
@@ -170,7 +170,7 @@ def mock_price_logs_raw_kafka():
                     'price': 163.87,
                     'sessionId': 'd423ce8a-77aa-4c9a-94d4-d1adddcc3472',
                     'experimentId': '53aefd07-f66a-4d7f-ba8b-7ea1fc562d35',
-                    'storeMode': 'shop',
+                    'storeMode': 'hotel',
                     'ts': '2025-11-25T21:05:58.009Z'
                 }
             }
@@ -183,7 +183,7 @@ def mock_price_logs_raw_kafka():
                     'price': 397.46,
                     'sessionId': 'd423ce8a-77aa-4c9a-94d4-d1adddcc3472',
                     'experimentId': '53aefd07-f66a-4d7f-ba8b-7ea1fc562d35',
-                    'storeMode': 'shop',
+                    'storeMode': 'hotel',
                     'ts': '2025-11-25T21:05:58.049Z'
                 }
             }
@@ -196,7 +196,7 @@ def mock_price_logs_raw_kafka():
                     'price': 401.66,
                     'sessionId': 'd423ce8a-77aa-4c9a-94d4-d1adddcc3472',
                     'experimentId': '53aefd07-f66a-4d7f-ba8b-7ea1fc562d35',
-                    'storeMode': 'shop',
+                    'storeMode': 'hotel',
                     'ts': '2025-11-25T21:06:08.864Z'
                 }
             }
@@ -222,7 +222,7 @@ def mock_experiments():
         'created_at': pd.to_datetime(['2025-11-25T20:00:00Z', '2025-11-26T10:00:00Z']),
         'subject_name': ['Session A', 'Session B'],
         'xp_human_only': [True, False],
-        'xp_market_mode': ['hotel', 'shop'],
+        'xp_market_mode': ['hotel', 'airline'],
         'xp_task_id': [None, None]
     })
 
@@ -269,3 +269,13 @@ def empty_context(empty_provider):
         store_mode='hotel',
         window_size='30s'
     )
+
+
+@pytest.fixture
+def session_interactions(mock_interactions):
+    """Enriched interaction data for session feature extraction tests"""
+    df = mock_interactions.copy()
+    df['userAgent'] = ['Mozilla/5.0 Chrome/120', 'Mozilla/5.0 Chrome/120',
+                       'HeadlessChrome/120', 'HeadlessChrome/120', 'HeadlessChrome/120']
+    df['metadata_base_price'] = [None, None, 150.0, 150.0, 200.0]
+    return df

experiments/procesing/tests/test_elasticity.py

@@ -1,353 +0,0 @@
-import pytest
-import pandas as pd
-import numpy as np
-from procesing.steps import (
-    AggregatePriceLogsStep,
-    ComputeElasticityStep
-)
-
-
-def test_aggregate_price_logs_basic(pipeline_context):
-    """Test basic price aggregation into time windows"""
-    step = AggregatePriceLogsStep(pipeline_context)
-
-    # Create price logs with known window structure
-    df = pd.DataFrame({
-        'ts': pd.date_range(start='2023-01-01 10:00:00', periods=100, freq='10s'),
-        'productId': np.tile([
-            'd018efc1-25e9-4284-b276-80386e048b25',
-            '51266ddb-5b07-47b7-89ee-5b5cae94bb11',
-            '2cd7f756-fc65-4ba0-ab01-74521c1fff43'
-        ], 34)[:100],
-        'price': np.random.uniform(100, 200, 100)
-    })
-
-    result = step.transform(df)
-    assert isinstance(result, list)
-    assert len(result) > 0
-    # each chunk should have window metadata and price vector
-    for chunk in result:
-        assert 'window_start' in chunk
-        assert 'window_end' in chunk
-        assert 'price_vector' in chunk
-        assert isinstance(chunk['price_vector'], pd.DataFrame)
-        assert 'productId' in chunk['price_vector'].columns
-        assert 'price' in chunk['price_vector'].columns
-
-
-def test_aggregate_price_logs_handles_gaps(pipeline_context):
-    """Test that price aggregation forward-fills missing windows"""
-    step = AggregatePriceLogsStep(pipeline_context)
-
-    # create sparse data with gaps
-    df = pd.DataFrame({
-        'ts': pd.to_datetime([
-            '2023-01-01 10:00:00',
-            '2023-01-01 10:00:05',
-            '2023-01-01 10:02:00',  # gap of ~2 mins
-            '2023-01-01 10:02:30'
-        ]),
-        'productId': [
-            'd018efc1-25e9-4284-b276-80386e048b25',
-            'd018efc1-25e9-4284-b276-80386e048b25',
-            '51266ddb-5b07-47b7-89ee-5b5cae94bb11',
-            '51266ddb-5b07-47b7-89ee-5b5cae94bb11'
-        ],
-        'price': [100, 102, 150, 153]
-    })
-
-    result = step.transform(df)
-    assert isinstance(result, list)
-    # should have multiple windows despite gaps
-    assert len(result) >= 2
-
-
-def test_compute_elasticity_with_known_relationship(pipeline_context):
-    """Test elasticity computation with known price-demand relationship"""
-    step = ComputeElasticityStep(pipeline_context)
-
-    # simulate elastic demand: when price ↑10%, demand ↓15% (elasticity ~ -1.5)
-    base_price = 100
-    base_demand = 50
-
-    demand_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'demand_score': [base_demand]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'demand_score': [base_demand * 0.85]  # 15% decrease
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:01:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:30'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'demand_score': [base_demand * 0.70]  # further decrease
-            })
-        }
-    ]
-
-    price_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'price_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'price': [base_price]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'price_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'price': [base_price * 1.10]  # 10% increase
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:01:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:30'),
-            'price_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'price': [base_price * 1.20]  # 20% increase
-            })
-        }
-    ]
-
-    result = step.transform((demand_chunks, price_chunks))
-    assert isinstance(result, pd.DataFrame)
-    assert not result.empty
-    assert 'productId' in result.columns
-    assert 'elasticity' in result.columns
-    assert 'n_obs' in result.columns
-
-    # check elasticity is negative (normal good)
-    product_elast = result[result['productId'] == 'd018efc1-25e9-4284-b276-80386e048b25']
-    assert len(product_elast) == 1
-    assert product_elast.iloc[0]['elasticity'] < 0
-    # should be roughly elastic (< -1)
-    assert product_elast.iloc[0]['n_obs'] == 3
-
-
-def test_compute_elasticity_inelastic_product(pipeline_context):
-    """Test with inelastic demand: price changes, demand barely moves"""
-    step = ComputeElasticityStep(pipeline_context)
-
-    base_price = 150
-    base_demand = 40
-
-    demand_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['51266ddb-5b07-47b7-89ee-5b5cae94bb11'],
-                'demand_score': [base_demand]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['51266ddb-5b07-47b7-89ee-5b5cae94bb11'],
-                'demand_score': [base_demand * 0.98]  # tiny 2% decrease
-            })
-        }
-    ]
-
-    price_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'price_vector': pd.DataFrame({
-                'productId': ['51266ddb-5b07-47b7-89ee-5b5cae94bb11'],
-                'price': [base_price]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'price_vector': pd.DataFrame({
-                'productId': ['51266ddb-5b07-47b7-89ee-5b5cae94bb11'],
-                'price': [base_price * 1.20]  # 20% increase
-            })
-        }
-    ]
-
-    result = step.transform((demand_chunks, price_chunks))
-    product_elast = result[result['productId'] == '51266ddb-5b07-47b7-89ee-5b5cae94bb11']
-    assert len(product_elast) == 1
-    # inelastic: elasticity between 0 and -1
-    assert -1 < product_elast.iloc[0]['elasticity'] < 0
-
-
-def test_compute_elasticity_multiple_products(pipeline_context):
-    """Test elasticity computation across multiple products simultaneously"""
-    step = ComputeElasticityStep(pipeline_context)
-
-    products = [
-        'd018efc1-25e9-4284-b276-80386e048b25',
-        '51266ddb-5b07-47b7-89ee-5b5cae94bb11',
-        '2cd7f756-fc65-4ba0-ab01-74521c1fff43'
-    ]
-
-    # create 5 time windows with all 3 products
-    demand_chunks = []
-    price_chunks = []
-
-    for i in range(5):
-        ts = pd.Timestamp('2023-01-01 10:00:00') + pd.Timedelta(f'{i*30}s')
-
-        demand_chunks.append({
-            'window_start': ts,
-            'window_end': ts + pd.Timedelta('30s'),
-            'demand_vector': pd.DataFrame({
-                'productId': products,
-                'demand_score': [
-                    50 * (0.9 ** i),  # elastic: decreases as price rises
-                    40 * (0.98 ** i), # inelastic: barely changes
-                    30 * (0.85 ** i)  # very elastic
-                ]
-            })
-        })
-
-        price_chunks.append({
-            'window_start': ts,
-            'window_end': ts + pd.Timedelta('30s'),
-            'price_vector': pd.DataFrame({
-                'productId': products,
-                'price': [
-                    100 * (1.05 ** i),
-                    150 * (1.10 ** i),
-                    120 * (1.08 ** i)
-                ]
-            })
-        })
-
-    result = step.transform((demand_chunks, price_chunks))
-    assert isinstance(result, pd.DataFrame)
-    assert len(result) == 3  # all products should have elasticity
-    assert set(result['productId']) == set(products)
-    assert all(result['n_obs'] == 5)
-    assert all(result['elasticity'] < 0)  # all normal goods
-
-
-def test_compute_elasticity_insufficient_data(pipeline_context):
-    """Test behavior with insufficient observations"""
-    step = ComputeElasticityStep(pipeline_context)
-
-    # only 1 observation
-    demand_chunks = [{
-        'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-        'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-        'demand_vector': pd.DataFrame({
-            'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-            'demand_score': [50]
-        })
-    }]
-
-    price_chunks = [{
-        'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-        'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-        'price_vector': pd.DataFrame({
-            'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-            'price': [100]
-        })
-    }]
-
-    result = step.transform((demand_chunks, price_chunks))
-    # should still return result but with low n_obs
-    product_elast = result[result['productId'] == 'd018efc1-25e9-4284-b276-80386e048b25']
-    assert len(product_elast) == 1
-    assert product_elast.iloc[0]['n_obs'] == 1
-    assert product_elast.iloc[0]['elasticity'] == 0.0  # not enough data
-
-
-def test_compute_elasticity_misaligned_chunks(pipeline_context):
-    """Test with non-overlapping demand and price windows"""
-    step = ComputeElasticityStep(pipeline_context)
-
-    demand_chunks = [{
-        'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-        'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-        'demand_vector': pd.DataFrame({
-            'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-            'demand_score': [50]
-        })
-    }]
-
-    price_chunks = [{
-        'window_start': pd.Timestamp('2023-01-01 11:00:00'),  # different time
-        'window_end': pd.Timestamp('2023-01-01 11:00:30'),
-        'price_vector': pd.DataFrame({
-            'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-            'price': [100]
-        })
-    }]
-
-    result = step.transform((demand_chunks, price_chunks))
-    # should handle gracefully with no aligned data
-    assert isinstance(result, pd.DataFrame)
-    assert all(result['n_obs'] == 0)
-
-
-def test_elasticity_arc_method(pipeline_context):
-    """Test arc elasticity computation method"""
-    # configure context for arc method
-    pipeline_context.config['elasticity_method'] = 'arc'
-    step = ComputeElasticityStep(pipeline_context)
-
-    demand_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'demand_score': [100]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'demand_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'demand_score': [80]
-            })
-        }
-    ]
-
-    price_chunks = [
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:00'),
-            'window_end': pd.Timestamp('2023-01-01 10:00:30'),
-            'price_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'price': [100]
-            })
-        },
-        {
-            'window_start': pd.Timestamp('2023-01-01 10:00:30'),
-            'window_end': pd.Timestamp('2023-01-01 10:01:00'),
-            'price_vector': pd.DataFrame({
-                'productId': ['d018efc1-25e9-4284-b276-80386e048b25'],
-                'price': [110]
-            })
-        }
-    ]
-
-    result = step.transform((demand_chunks, price_chunks))
-    product_elast = result[result['productId'] == 'd018efc1-25e9-4284-b276-80386e048b25']
-    assert len(product_elast) == 1
-    assert product_elast.iloc[0]['elasticity'] < 0
-    # reset config
-    pipeline_context.config['elasticity_method'] = 'point'

lib/model_registry.py

@@ -26,6 +26,7 @@ class ModelRegistry:
         self.metadata_prefix = "model:meta:"
         self.data_prefix = "model:data:"
         self.elasticity_prefix = "elasticity:"
+        self.prices_prefix = "prices:"
 
     def publish_elasticity(self,
                           elasticity_df: pd.DataFrame,
@@ -130,6 +131,46 @@ class ModelRegistry:
 
         return models
 
+    def publish_prices(self,
+                      prices_df: pd.DataFrame,
+                      model_name: str = 'latest',
+                      metadata: Optional[Dict[str, Any]] = None):
+        """Store predicted prices in registry.
+
+        Args:
+            prices_df: df with [productId, predicted_price, ...]
+            model_name: identifier for this price snapshot
+            metadata: additional info
+        """
+        key = f"{self.prices_prefix}{model_name}"
+        data_json = prices_df.to_json(orient='records')
+
+        self.redis_client.set(key, data_json)
+
+        meta = metadata or {}
+        meta.update({
+            'n_products': len(prices_df),
+            'model_type': 'predicted_prices'
+        })
+
+        meta_key = f"{self.metadata_prefix}prices_{model_name}"
+        self.redis_client.set(meta_key, json.dumps(meta))
+
+        log.info(f"Published prices '{model_name}' for {len(prices_df)} products")
+
+    def get_prices(self, model_name: str = 'latest') -> Optional[pd.DataFrame]:
+        """Retrieve predicted prices from registry."""
+        key = f"{self.prices_prefix}{model_name}"
+        data_json = self.redis_client.get(key)
+
+        if data_json is None:
+            return None
+
+        if isinstance(data_json, bytes):
+            data_json = data_json.decode('utf-8')
+
+        return pd.read_json(data_json, orient='records')
+
     def health_check(self) -> bool:
         """Check if Redis connection is alive."""
         try:
@@ -137,3 +178,49 @@ class ModelRegistry:
             return True
         except:
             return False
+
+    def set_session_prices(self, session_id: str, prices: Dict[str, float], ttl: int = 1800):
+        """
+        Store prices for a specific session.
+        THIS is the write path for session-aware pricing.
+
+        Args:
+            session_id: session identifier
+            prices: dict of {productId: price}
+            ttl: time-to-live in seconds (default 30min)
+        """
+        if not prices:
+            return
+
+        key = f"session:{session_id}:prices"
+        # use Redis hash for O(1) lookup per product
+        self.redis_client.hset(key, mapping={k: str(v) for k, v in prices.items()})
+        self.redis_client.expire(key, ttl)
+
+    def get_session_price(self, session_id: str, product_id: str) -> Optional[float]:
+        """
+        Lookup price for (sessionId, productId).
+        THIS is the read path for fast provider lookup.
+
+        Returns: price or None if not found
+        """
+        key = f"session:{session_id}:prices"
+        price_str = self.redis_client.hget(key, product_id)
+
+        if price_str is None:
+            return None
+
+        return float(price_str.decode('utf-8') if isinstance(price_str, bytes) else price_str)
+
+    def get_session_all_prices(self, session_id: str) -> Dict[str, float]:
+        """Get all prices for a session."""
+        key = f"session:{session_id}:prices"
+        prices_raw = self.redis_client.hgetall(key)
+
+        if not prices_raw:
+            return {}
+
+        return {
+            (k.decode('utf-8') if isinstance(k, bytes) else k): float(v.decode('utf-8') if isinstance(v, bytes) else v)
+            for k, v in prices_raw.items()
+        }

requirements.txt

@@ -11,3 +11,4 @@ pytest-asyncio
 uv
 scikit-learn
 supabase
+pymc

sim/rl/behavior_loader/loader.py

@@ -0,0 +1,63 @@
+import os
+from pydantic import BaseModel as Base
+import json
+
+class PayloadModel(Base):
+    sessionId: str
+    experimentId: str | None
+    eventName: str
+    page: str | None
+    productId: str | None
+    metadata: dict
+    storeMode: str
+    userAgent: str
+    ts: str
+
+class ValueModel(Base):
+    payload: PayloadModel
+    encoding: str
+    isPayloadNull: bool
+    schemaId: int
+    size: int
+
+class InteractionModel(Base):
+    partitionID: int
+    offset: int
+    timestamp: int
+    compression: str
+    isTransactional: bool
+    headers: list
+    key: dict
+    value: ValueModel
+
+class Loader:
+    def __init__(self, src_dir: str):
+        self.src_dir = src_dir
+        self.entries = os.listdir(src_dir)
+        if not self.entries: raise ValueError("empty directory")
+        self.data = self._load_sessions()
+
+    def _is_admin_page(self, interaction: InteractionModel) -> bool:
+        page = interaction.value.payload.page
+        return page and page.startswith("/admin/")
+
+    def _load_sessions(self) -> dict:
+        sessions = {}
+        for entry in self.entries:
+            int_path = f"{self.src_dir}/{entry}/int.json"
+            raw = json.load(open(int_path))
+            ints = [InteractionModel(**i) for i in raw]
+            sessions[entry] = [i for i in ints if not self._is_admin_page(i)]
+        return sessions
+
+    def get_data(self) -> dict:
+        return self.data
+
+    def get_entries(self) -> tuple[list[str], int]:
+        return self.entries, len(self.entries)
+
+if __name__ == "__main__":
+    DIR = "/home/velocitatem/Documents/Projects/PHANTOM/experiments/collected_data/"
+    loader = Loader(DIR)
+    _, n = loader.get_entries()
+    print(f"Loaded {n} sessions from {DIR}")

sim/rl/behavior_loader/models.py

@@ -0,0 +1,144 @@
+from loader import Loader
+from collections import defaultdict
+from typing import Dict, List, Tuple, Set
+import numpy as np
+import graphviz
+
+DIR = "/home/velocitatem/Documents/Projects/PHANTOM/experiments/collected_data/"
+
+class BehaviorModel:
+    def __init__(self, src_dir: str = DIR):
+        self.loader = Loader(src_dir)
+        self.data = self.loader.get_data()
+        self.entries, self.num_entries = self.loader.get_entries()
+        self.mdp = None
+
+    def _state_repr(self, evt) -> str:
+        p = evt.value.payload
+        return f"{p.page or 'unk'}|{p.productId or 'none'}|{p.eventName}"
+
+    def _extract_sessions(self):
+        # transform raw events into sequential state trajectories per session
+        trajectories = []
+        for sid, evts in self.data.items():
+            if len(evts) < 2: continue
+            states = [self._state_repr(e) for e in sorted(evts, key=lambda x: x.timestamp)]
+            trajectories.append(states)
+        return trajectories
+
+    def _calc_transitions(self, trajectories: List[List[str]]) -> Tuple[Dict, Set]:
+        trans = defaultdict(lambda: defaultdict(int))
+        states = set()
+        for traj in trajectories:
+            for i in range(len(traj) - 1):
+                s, s_next = traj[i], traj[i+1]
+                trans[s][s_next] += 1
+                states.update([s, s_next])
+        return trans, states
+
+    def _calc_rewards(self, trajectories: List[List[str]]) -> Dict:
+        # reward based on session progression depth
+        rwd = defaultdict(list)
+        for traj in trajectories:
+            n = len(traj)
+            for i, s in enumerate(traj):
+                rwd[s].append(i / n)
+        return rwd
+
+    def _normalize_trans(self, counts: Dict) -> Dict:
+        return {s: {s_n: cnt/sum(nxt.values()) for s_n, cnt in nxt.items()}
+                for s, nxt in counts.items()}
+
+    def build_MDP(self) -> Dict:
+        trajs = self._extract_sessions()
+        trans_cnt, states = self._calc_transitions(trajs)
+        trans_prob = self._normalize_trans(trans_cnt)
+        state_rwd = self._calc_rewards(trajs)
+        state_val = {s: np.mean(r) for s, r in state_rwd.items()}
+
+        self.mdp = {
+            'states': sorted(list(states)),
+            'num_states': len(states),
+            'transitions': trans_prob,
+            'state_values': state_val,
+            'state_rewards': state_rwd,
+            'trans_counts': trans_cnt,
+        }
+        return self.mdp
+
+    def transition_prob(self, s: str, s_next: str) -> float:
+        if not self.mdp: raise ValueError("build MDP first")
+        return self.mdp['transitions'].get(s, {}).get(s_next, 0.0)
+
+    def state_value(self, s: str) -> float:
+        if not self.mdp: raise ValueError("build MDP first")
+        return self.mdp['state_values'].get(s, 0.0)
+
+    def sample_traj(self, start: str, max_len: int = 50) -> List[str]:
+        if not self.mdp: raise ValueError("build MDP first")
+        path = [start]
+        curr = start
+        for _ in range(max_len):
+            nxt = self.mdp['transitions'].get(curr, {})
+            if not nxt: break
+            curr = np.random.choice(list(nxt.keys()), p=list(nxt.values()))
+            path.append(curr)
+        return path
+
+def visualize_mdp(model: BehaviorModel, threshold: float = 0.05, output: str = "mdp_graph", fmt: str = "svg", view: bool = False, export_dot: bool = False):
+    """visualize MDP as directed graph using graphviz, aggregated by event type"""
+    if not model.mdp: raise ValueError("build MDP first")
+
+    # aggregate transitions by event type
+    evt_trans = defaultdict(lambda: defaultdict(float))
+    for s, trans in model.mdp['transitions'].items():
+        evt_src = s.split('|')[2]
+        for s_next, prob in trans.items():
+            evt_dst = s_next.split('|')[2]
+            evt_trans[evt_src][evt_dst] += prob
+
+    # normalize aggregated transitions
+    for evt_src in evt_trans:
+        total = sum(evt_trans[evt_src].values())
+        if total > 0:
+            for evt_dst in evt_trans[evt_src]:
+                evt_trans[evt_src][evt_dst] /= total
+
+    g = graphviz.Digraph(format=fmt)
+    g.attr(rankdir='LR', size='30')
+    g.attr('node', shape='circle', width='1', height='1')
+
+    # collect all event types
+    events = set(evt_trans.keys())
+    for trans in evt_trans.values():
+        events.update(trans.keys())
+
+    # add nodes for each event type
+    for evt in events:
+        g.node(evt)
+
+    # add edges above threshold
+    for evt_src in evt_trans:
+        for evt_dst, prob in evt_trans[evt_src].items():
+            if prob > threshold:
+                g.edge(evt_src, evt_dst, label=f'{prob:.2f}')
+
+    g.render(output, view=view, cleanup=True)
+    print(f"Saved MDP graph to {output}.{fmt}")
+
+    if export_dot:
+        dot_file = f"{output}.dot"
+        with open(dot_file, 'w') as f:
+            f.write(g.source)
+        print(f"Exported DOT source to {dot_file}")
+
+    return g
+
+if __name__ == "__main__":
+    model = BehaviorModel(DIR)
+    mdp = model.build_MDP()
+    print(f"Built MDP: {mdp['num_states']} states, {sum(len(t) for t in mdp['transitions'].values())} transitions")
+    if not mdp['states']:
+        print("No states found")
+        exit(1)
+    visualize_mdp(model, threshold=0.05, output="mdp_viz", fmt="pdf", export_dot=True)

sim/rl/engine.py

@@ -0,0 +1,227 @@
+from os import kill
+import numpy as np
+import pandas as pd
+from abc import ABC, abstractmethod
+from typing import Dict, Any
+from environment import BusinessLogicConstraints
+
+
+"""
+An angine by default should have its own demand estimation mechanism from the observed observations whihc are the computer feature.
+From these features we then follow the researc hstructure of q -> p with a testable and must be updatable mechanism.
+"""
+
+class BasePricingEngine(ABC):
+    """base interface for all pricing engines"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        self.c = constraints
+        self.rng = np.random.default_rng(seed)
+        self.step_count = 0
+
+
+    @abstractmethod
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        """compute new prices given current state and observation from environment
+
+        args:
+            current_prices: current price vector [N]
+            observation: dict containing 'price', 'demand', and possibly interaction data
+
+        returns:
+            new_prices: updated price vector [N]
+        """
+        pass
+
+    @abstractmethod
+    def update(obs, reward, done, info):
+        pass
+
+
+
+    def reset(self):
+        """reset engine state for new episode"""
+        self.step_count = 0
+
+
+class WildPricingEngine(BasePricingEngine):
+    """production-like pricing using online elasticity estimation via EWMA regression"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        super().__init__(constraints, seed)
+        # per-product unit costs (unknown to customers; known to platform)
+        self.unit_cost = self.rng.uniform(8.0, 40.0, size=self.c.product_catelogue_size).astype(np.float32)
+        # online elasticity estimate (start moderately elastic)
+        self.e_hat = np.full((self.c.product_catelogue_size,), -1.3, dtype=np.float32)
+        # EWMA state for log-log regression
+        self.mu_logp = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.mu_logq = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.cov_pq  = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.var_p   = np.ones(self.c.product_catelogue_size, dtype=np.float32)
+        # knobs typical in production
+        self.lr = 0.08
+        self.ewma = 0.05
+        self.eps_explore = 0.03
+        self.explore_scale = 0.03
+
+    def _safe_elasticity(self, e: np.ndarray) -> np.ndarray:
+        return np.clip(e, -5.0, -1.05)
+
+    def reset(self):
+        super().reset()
+        self.e_hat = np.full((self.c.product_catelogue_size,), -1.3, dtype=np.float32)
+        self.mu_logp = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.mu_logq = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.cov_pq = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        self.var_p = np.ones(self.c.product_catelogue_size, dtype=np.float32)
+
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        self.step_count += 1
+        # extract demand signal (from env observation) as proxy for sales
+        demand = observation.get('demand', np.zeros(self.c.product_catelogue_size, dtype=np.float32))
+        return self._update_from_demand(current_prices, demand)
+
+    def _update_from_demand(self, prices: np.ndarray, sold: np.ndarray) -> np.ndarray:
+        # log transforms (add 1 to handle zeros)
+        logp = np.log(np.clip(prices, 1e-3, None)).astype(np.float32)
+        logq = np.log(sold + 1.0).astype(np.float32)
+        # EWMA moments for per-product regression: logq ≈ a + e*logp
+        a = self.ewma
+        dp = logp - self.mu_logp
+        dq = logq - self.mu_logq
+        self.mu_logp = (1 - a) * self.mu_logp + a * logp
+        self.mu_logq = (1 - a) * self.mu_logq + a * logq
+        self.cov_pq = (1 - a) * self.cov_pq + a * (dp * dq)
+        self.var_p = (1 - a) * self.var_p + a * (dp * dp + 1e-6)
+        e_new = self.cov_pq / (self.var_p + 1e-6)
+        self.e_hat = self._safe_elasticity(0.9 * self.e_hat + 0.1 * e_new)
+        # profit-optimal price for isoelastic demand (if e < -1)
+        e = self.e_hat
+        p_star = self.unit_cost * (e / (e + 1.0))
+        # smooth toward p_star
+        new_prices = (1 - self.lr) * prices + self.lr * p_star
+        # exploration (small random perturbations)
+        if self.rng.random() < self.eps_explore:
+            noise = self.rng.normal(0.0, self.explore_scale, size=new_prices.shape).astype(np.float32)
+            new_prices = new_prices * (1.0 + noise)
+        # apply business guardrails (max change + bounds)
+        max_adj = self.c.max_price_adjustment
+        ratio = np.clip(new_prices / (prices + 1e-6), 1 - max_adj, 1 + max_adj)
+        new_prices = prices * ratio
+        new_prices = np.clip(new_prices, self.c.system_min_price, self.c.system_max_price).astype(np.float32)
+        return new_prices
+
+
+class StaticPricingEngine(BasePricingEngine):
+    """baseline: fixed prices throughout episode"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        super().__init__(constraints, seed)
+        self.fixed_prices = None
+
+    def reset(self):
+        super().reset()
+        self.fixed_prices = None
+
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        self.step_count += 1
+        if self.fixed_prices is None:
+            self.fixed_prices = current_prices.copy()
+        return self.fixed_prices.copy()
+
+
+class SimpleDemandEngine(BasePricingEngine):
+    """demand-driven pricing: increase price when demand rises, decrease when it falls"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        super().__init__(constraints, seed)
+        self.prev_demand = None
+        self.lr = 0.05
+
+    def reset(self):
+        super().reset()
+        self.prev_demand = None
+
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        self.step_count += 1
+        demand = observation.get('demand', np.zeros(self.c.product_catelogue_size, dtype=np.float32))
+        if self.prev_demand is None:
+            self.prev_demand = demand.copy()
+            return current_prices.copy()
+        # simple rule: if demand increases, raise price; if decreases, lower price
+        delta_d = demand - self.prev_demand
+        price_adj = self.lr * np.sign(delta_d) * np.abs(delta_d) / (np.abs(self.prev_demand) + 1.0)
+        new_prices = current_prices * (1.0 + price_adj)
+        self.prev_demand = demand.copy()
+        # apply constraints
+        max_adj = self.c.max_price_adjustment
+        ratio = np.clip(new_prices / (current_prices + 1e-6), 1 - max_adj, 1 + max_adj)
+        new_prices = current_prices * ratio
+        return np.clip(new_prices, self.c.system_min_price, self.c.system_max_price).astype(np.float32)
+
+
+class RandomWalkEngine(BasePricingEngine):
+    """random walk pricing with mean reversion"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        super().__init__(constraints, seed)
+        self.target_price = None
+        self.volatility = 0.02
+
+    def reset(self):
+        super().reset()
+        self.target_price = None
+
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        self.step_count += 1
+        if self.target_price is None:
+            self.target_price = current_prices.copy()
+        # random walk with mean reversion toward target
+        noise = self.rng.normal(0.0, self.volatility, size=current_prices.shape).astype(np.float32)
+        reversion = 0.01 * (self.target_price - current_prices)
+        new_prices = current_prices * (1.0 + noise) + reversion
+        # apply constraints
+        max_adj = self.c.max_price_adjustment
+        ratio = np.clip(new_prices / (current_prices + 1e-6), 1 - max_adj, 1 + max_adj)
+        new_prices = current_prices * ratio
+        return np.clip(new_prices, self.c.system_min_price, self.c.system_max_price).astype(np.float32)
+
+
+class ThompsonSamplingEngine(BasePricingEngine):
+    """bayesian bandit approach per product treating price as discrete action"""
+    def __init__(self, constraints: BusinessLogicConstraints, seed: int = 0):
+        super().__init__(constraints, seed)
+        self.n_price_levels = 5
+        self.alpha = np.ones((self.c.product_catelogue_size, self.n_price_levels), dtype=np.float32)
+        self.beta = np.ones((self.c.product_catelogue_size, self.n_price_levels), dtype=np.float32)
+        self.price_grid = None
+        self.last_actions = None
+
+    def reset(self):
+        super().reset()
+        self.alpha = np.ones((self.c.product_catelogue_size, self.n_price_levels), dtype=np.float32)
+        self.beta = np.ones((self.c.product_catelogue_size, self.n_price_levels), dtype=np.float32)
+        self.price_grid = None
+        self.last_actions = None
+
+    def compute_prices(self, current_prices: np.ndarray, observation: Dict[str, Any]) -> np.ndarray:
+        self.step_count += 1
+        if self.price_grid is None:
+            # define price grid per product
+            lo = current_prices * 0.7
+            hi = current_prices * 1.3
+            self.price_grid = np.linspace(lo, hi, self.n_price_levels).T
+        demand = observation.get('demand', np.zeros(self.c.product_catelogue_size, dtype=np.float32))
+        # update beliefs based on last action
+        if self.last_actions is not None:
+            for i in range(self.c.product_catelogue_size):
+                a = self.last_actions[i]
+                reward = demand[i]
+                if reward > 0.5:
+                    self.alpha[i, a] += reward
+                else:
+                    self.beta[i, a] += 1.0
+        # thompson sampling: sample from posterior, pick best
+        new_prices = np.zeros(self.c.product_catelogue_size, dtype=np.float32)
+        actions = np.zeros(self.c.product_catelogue_size, dtype=int)
+        for i in range(self.c.product_catelogue_size):
+            theta = self.rng.beta(self.alpha[i], self.beta[i]).astype(np.float32)
+            actions[i] = int(np.argmax(theta))
+            new_prices[i] = self.price_grid[i, actions[i]]
+        self.last_actions = actions
+        return np.clip(new_prices, self.c.system_min_price, self.c.system_max_price).astype(np.float32)

sim/rl/environment.py

@@ -0,0 +1,320 @@
+from sys import intern
+import gymnasium as gym
+from gymnasium import spaces
+from matplotlib import interactive
+import numpy as np
+from dataclasses import dataclass
+import pandas as pd
+from typing import Callable, Optional, Dict, Any, List
+
+# "learner"  agent learning to optimize pricing
+# "agent"  part of environment creating demand signals that learner processes
+
+@dataclass
+class BusinessLogicConstraints():
+    max_price_adjustment: float = 0.30
+    system_max_price: float = 500.0
+    system_min_price: float = 1.0
+    product_catelogue_size: int = 100
+    episode_length: int = 200
+    sessions_per_step: int = 250
+    agent_share: float = 0.25
+    agent_recon_multiplier: float = 6.0
+    agent_purchase_probability: float = 0.20
+    coi_strength: float = 0.25
+    coi_threshold: float = 4.0
+    coi_sigmoid_temp: float = 1.25
+    base_human_demand: float = 0.08
+    base_agent_demand: float = 0.05
+    human_price_elasticity: float = -1.2 # assumptions here
+    agent_price_elasticity: float = -0.6
+    w_agent_loss: float = 1.0
+    w_volatility: float = 5.0
+    w_estimation_error: float = 0.25
+    seed: int = 7
+
+
+def _sigmoid(x: np.ndarray) -> np.ndarray:
+    return 1.0 / (1.0 + np.exp(-x))
+
+class CommercePlatform:
+    """
+    This is just an extension of the state management for the environment, it does not implement anything dynamic just helps us simulate demand.
+    """
+    def __init__(self,
+                 product_catelogue_size: int,
+                 max_price: float,
+                 min_price: float,
+                 constraints: BusinessLogicConstraints):
+        self.product_catelogue_size = product_catelogue_size
+        self.product_supply = np.random.uniform(low=10, high=50, size=(self.product_catelogue_size,))
+        self.max_price = max_price
+        self.min_price = min_price
+        self.constraints = constraints
+        self.simulation_history: List[Dict[str, Any]] = []
+        self._rng = np.random.default_rng(constraints.seed)
+        self._last_interaction_df: pd.DataFrame = pd.DataFrame()
+
+
+    def setup_true_demand(self, prices: np.ndarray) -> Dict[str, np.ndarray]:
+        # ground truth purchase propensities
+        p = np.clip(prices, self.min_price, self.max_price)
+        pn = p / self.max_price
+        human_prob = self.constraints.base_human_demand * (pn ** self.constraints.human_price_elasticity)
+        agent_prob = self.constraints.base_agent_demand * (pn ** self.constraints.agent_price_elasticity)
+        return {
+            "human_purchase_prob": np.clip(human_prob, 0.0, 0.95),
+            "agent_purchase_prob": np.clip(agent_prob, 0.0, 0.95)
+        }
+
+    def _load_behavioral_profile(actor : str, demand_forcing):
+        """
+        This returns a markov chain with average weights which we get from interaction data of our experiments.
+        This defines transition probabilities between different events:
+        search -> view_item_price_binN: 0.7
+        view_item_price_binN -> add_to_cart: 0.2
+        we also must reweight with the demand_forcing vector or purchase probabilities per-product
+        """
+
+
+    def _simulate_sessions(self, base_prices: np.ndarray) -> pd.DataFrame:
+        demand = self.setup_true_demand(base_prices)
+        human_pprob = demand["human_purchase_prob"]
+        agent_pprob = demand["agent_purchase_prob"]
+        events: List[Dict[str, Any]] = []
+        T = self.constraints.sessions_per_step
+        n_agent_sessions = int(round(T * self.constraints.agent_share))
+        n_human_sessions = T - n_agent_sessions
+        n_agent_ids = max(1, n_agent_sessions // 2)
+        session_map = {
+            'humans': n_human_sessions,
+            'agents': n_agent_ids
+        }
+        pprob_map = {
+            'humans': human_pprob,
+            'agents': agent_pprob
+        }
+        joint_events = []
+        for actor, n_sessions in session_map.items():
+            bp = _load_behavioral_profile(actor, pprob_map[actor])
+            counter = 0
+            events = []
+            while counter < n_sessions:
+                session_events = []
+                while len(session_events) == 0 or session_events[-1]['action'] == 'checkout':
+                    interaction_event = bp.sample(self._rng)
+                    interaction_event['session_id'] = f'{actor}_{counter:06d}'
+                    # TODO any other assignments
+                    session_events.append(interaction_event)
+                events.extend(session_events)
+                counter += 1
+            joint_events.extend(events)
+
+        return pd.DataFrame(joint_events)
+
+    def compute_interaction_features(self, interaction_df: pd.DataFrame) -> Dict[str, float]:
+        if interaction_df.empty:
+            return {"mean_sale_price": 0.0, "look_to_book": 0.0}
+        purchases = interaction_df[interaction_df["action"] == "purchase"]
+        mean_sale_price = float(purchases["price_paid"].mean()) if not purchases.empty else 0.0
+        views = float((interaction_df["action"] == "view").sum())
+        buys = float((interaction_df["action"] == "purchase").sum())
+        return {"mean_sale_price": mean_sale_price, "look_to_book": float(views / (buys + 1e-6))}
+
+    def _session_feature_table(self, df: pd.DataFrame) -> pd.DataFrame:
+        # TODO: adapt this
+        if df.empty:
+            return pd.DataFrame()
+        g = df.groupby("session_id", sort=False)
+        session_duration = g["t"].max() - g["t"].min()
+        total_interactions = g.size()
+        avg_time_between = g["t"].apply(lambda x: float(np.diff(np.sort(x.to_numpy())).mean()) if len(x) > 1 else 0.0)
+        interaction_velocity = total_interactions / (session_duration + 1e-6)
+        views = g.apply(lambda x: int((x["action"] == "view").sum()), include_groups=False)
+        cart_adds = g.apply(lambda x: int((x["action"] == "cart").sum()), include_groups=False)
+        purchases = g.apply(lambda x: int((x["action"] == "purchase").sum()), include_groups=False)
+        conversion_rate = purchases / (views + 1e-6)
+        is_agent = g["actor"].apply(lambda s: bool((s == "agent").any()), include_groups=False)
+
+        return pd.DataFrame({
+            "session_duration_sec": session_duration.astype(float),
+            "avg_time_between_events": avg_time_between.astype(float),
+            "total_interactions": total_interactions.astype(int),
+            "interaction_velocity": interaction_velocity.astype(float),
+            "item_views": views.astype(int),
+            "cart_adds": cart_adds.astype(int),
+            "purchases": purchases.astype(int),
+            "conversion_rate": conversion_rate.astype(float),
+            "is_agent": is_agent.astype(bool),
+        }).reset_index()
+
+    def get_interaction_data(self) -> np.ndarray:
+        if self._last_interaction_df.empty:
+            return np.array([], dtype=object)
+        return self._last_interaction_df.to_dict(orient="records")
+
+
+class PHANTOMEnv(gym.Env):
+    metadata = {"render_modes": []}
+
+    def __init__(self, constraints):
+        super().__init__()
+        self.constraints = BusinessLogicConstraints()
+        self.action_space = spaces.Box(low=-self.constraints.max_price_adjustment,
+                                       high=self.constraints.max_price_adjustment,
+                                       shape=(self.constraints.product_catelogue_size,), dtype=np.float32)
+        self.observation_space = spaces.Dict({
+            "elasticity": spaces.Dict({
+                "price": spaces.Box(
+                    low=np.full((self.constraints.product_catelogue_size,), self.constraints.system_min_price, dtype=np.float32),
+                    high=np.full((self.constraints.product_catelogue_size,), self.constraints.system_max_price, dtype=np.float32),
+                    dtype=np.float32),
+                "demand": spaces.Box(
+                    low=np.zeros((self.constraints.product_catelogue_size,), dtype=np.float32),
+                    high=np.full((self.constraints.product_catelogue_size,), 1e6, dtype=np.float32),
+                    dtype=np.float32),
+            })
+            # TODO: define more features that we compute from the interaction data
+        })
+        self.commerce_platform = CommercePlatform(
+            product_catelogue_size=self.constraints.product_catelogue_size,
+            max_price=self.constraints.system_max_price,
+            min_price=self.constraints.system_min_price,
+            constraints=self.constraints)
+        self._rng = np.random.default_rng(self.constraints.seed)
+        self.t = 0
+        self._prev_prices: Optional[np.ndarray] = None
+        self.state: Dict[str, Any] = {}
+
+    def reset(self, seed: Optional[int] = None, options: Optional[dict] = None):
+        super().reset(seed=seed)
+        if seed is not None:
+            self._rng = np.random.default_rng(seed)
+            self.commerce_platform._rng = np.random.default_rng(seed)
+        self.t = 0
+        init_prices = self._rng.uniform(low=60.0, high=140.0, size=(self.constraints.product_catelogue_size,)).astype(np.float32)
+        self._prev_prices = init_prices.copy()
+        self.state = {
+            "elasticity": {
+                "price": init_prices,
+                "demand": np.zeros((self.constraints.product_catelogue_size,), dtype=np.float32),
+            }
+        }
+        return self.state, {}
+
+    def step(self, action: np.ndarray):
+        self.t += 1
+        base_prices = self.state["elasticity"]["price"].astype(np.float32)
+        new_prices = np.clip(base_prices * (1.0 + action.astype(np.float32)),
+                           self.constraints.system_min_price,
+                           self.constraints.system_max_price).astype(np.float32)
+
+        self.state["elasticity"]["price"] = new_prices
+        # TODO: use the commerce platform to simulate sessions
+        interactions_df = self.commerce_platform._simulate_sessions(new_prices)
+        result = self.commerce_platform.compute_interaction_features(interactions_df)
+        # TODO: implement COI computation to use in reward
+        COI = 0.0
+
+        volatility = 0.0 if self._prev_prices is None else \
+            float(np.mean(np.abs((new_prices - self._prev_prices) / (self._prev_prices + 1e-6))))
+        self._prev_prices = new_prices.copy()
+
+        revenue_observed = float(result["revenue_observed"])
+        agent_loss = float(result["agent_loss"])
+
+        reward = (revenue_observed
+                  - COI
+                  - self.constraints.w_agent_loss * agent_loss
+                  - self.constraints.w_volatility * volatility
+                  - self.constraints.w_estimation_error
+                  )
+
+        terminated = self.t >= self.constraints.episode_length
+        info = {
+            "t": self.t,
+            "revenue_observed": revenue_observed,
+            "revenue_oracle": float(result["revenue_oracle"]),
+            "agent_loss": agent_loss,
+            "ux_volatility": volatility,
+            "mean_internal_error": err_mean,
+            "look_to_book": float(result["interaction_features"].get("look_to_book", 0.0)),
+            "mean_sale_price": float(result["interaction_features"].get("mean_sale_price", 0.0)),
+            "true_human_purchases_total": float(np.sum(result["true_human_demand"])),
+            "true_agent_purchases_total": float(np.sum(result["true_agent_purchases"])),
+        }
+        return self.state, float(reward), terminated, False, info
+
+
+if __name__ == "__main__":
+    import matplotlib.pyplot as plt
+    from collections import defaultdict
+
+    runs = {}
+    for use_defense in (False, True):
+        env = PHANTOMEnv(use_defense=use_defense)
+        obs, _ = env.reset(seed=42)
+        metrics = defaultdict(list)
+        total_reward = 0.0
+        done = False
+
+        while not done:
+            action = env.action_space.sample()
+            obs, reward, done, _, info = env.step(action)
+            total_reward += reward
+            p_mean = float(np.mean(obs["elasticity"]["price"]))
+            q_mean = float(np.mean(obs["elasticity"]["demand"]))
+            p_std = float(np.std(obs["elasticity"]["price"]))
+
+            metrics['t'].append(info['t'])
+            metrics['price_mean'].append(p_mean)
+            metrics['price_std'].append(p_std)
+            metrics['demand_mean'].append(q_mean)
+            metrics['revenue_observed'].append(info['revenue_observed'])
+            metrics['revenue_oracle'].append(info['revenue_oracle'])
+            metrics['agent_loss'].append(info['agent_loss'])
+            metrics['ux_volatility'].append(info['ux_volatility'])
+            metrics['look_to_book'].append(info['look_to_book'])
+            metrics['reward'].append(reward)
+            metrics['human_purchases'].append(info['true_human_purchases_total'])
+            metrics['agent_purchases'].append(info['true_agent_purchases_total'])
+
+            if info['t'] % 20 == 0 or done:
+                print(f"defense={'ON ' if use_defense else 'OFF'} t={info['t']:03d} p={p_mean:6.2f}±{p_std:4.2f} "
+                      f"q={q_mean:6.2f} rev={info['revenue_observed']:7.2f} oracle={info['revenue_oracle']:7.2f} "
+                      f"loss={info['agent_loss']:6.2f} ux={info['ux_volatility']:.3f} "
+                      f"ltb={info['look_to_book']:5.2f} r={reward:7.2f}")
+
+        runs[use_defense] = metrics
+        print(f"defense={'ON ' if use_defense else 'OFF'} total_reward={total_reward:.2f}\n")
+
+    fig, axes = plt.subplots(3, 3, figsize=(15, 12))
+    fig.suptitle('PHANTOM Environment: Defense OFF vs ON', fontsize=14, fontweight='bold')
+
+    plot_configs = [
+        ('price_mean', 'Mean Price', 'Price'),
+        ('demand_mean', 'Mean Demand Estimate', 'Demand'),
+        ('revenue_observed', 'Revenue (Observed)', 'Revenue'),
+        ('agent_loss', 'Agent Loss (Oracle - Observed)', 'Loss'),
+        ('ux_volatility', 'UX Volatility (Price Change)', 'Volatility'),
+        ('look_to_book', 'Look-to-Book Ratio', 'Ratio'),
+        ('reward', 'Step Reward', 'Reward'),
+        ('human_purchases', 'Human Purchases', 'Count'),
+        ('agent_purchases', 'Agent Purchases', 'Count'),
+    ]
+
+    for idx, (key, title, ylabel) in enumerate(plot_configs):
+        ax = axes[idx // 3, idx % 3]
+        for use_defense, label, color in [(False, 'No Defense', 'red'), (True, 'With Defense', 'blue')]:
+            m = runs[use_defense]
+            ax.plot(m['t'], m[key], label=label, color=color, alpha=0.7, linewidth=1.5)
+        ax.set_xlabel('Step')
+        ax.set_ylabel(ylabel)
+        ax.set_title(title, fontsize=10, fontweight='bold')
+        ax.legend(loc='best', fontsize=8)
+        ax.grid(True, alpha=0.3)
+
+    plt.tight_layout()
+    plt.savefig('phantom_env_comparison.png', dpi=150, bbox_inches='tight')
+    print("Plot saved to phantom_env_comparison.png")
+    plt.show()

sim/rl/train.py

@@ -0,0 +1,149 @@
+import numpy as np
+import logging
+from pathlib import Path
+from typing import Dict, Type, Optional
+import pickle
+from torch import neg_
+from torch.utils.tensorboard import SummaryWriter
+from environment import PHANTOMEnv, FastTrainingConstraints, BusinessLogicConstraints
+from engine import (BasePricingEngine, WildPricingEngine, StaticPricingEngine,
+                   SimpleDemandEngine, RandomWalkEngine, ThompsonSamplingEngine)
+
+logging.basicConfig(level=logging.INFO, format='%(asctime)s [%(levelname)s] %(message)s')
+logger = logging.getLogger(__name__)
+
+
+
+"""
+Target training loop:
+have base prices p0 from env reset and run the env step, collect reward and metrics
+pass this to the pricing engine which computes the price action to take based on previous reward by learning
+the new action gets passed to the step
+so we alternate, step -> reward -> engine (produces price delta) -> step with price delta -> reward
+to make sure the reinforcement learning inside the engine can learn we need to have trajectory of prices
+CURRENT SOLUTION BELOW does not implement correct learning or updates.
+"""
+
+class EngineTrainer:
+    """wrapper to run pricing engines through episodes and collect metrics"""
+    def __init__(self, engine: BasePricingEngine, env: PHANTOMEnv,
+                 tb_writer: Optional[SummaryWriter] = None):
+        self.engine = engine
+        self.env = env
+        self.episode_metrics = []
+        self.tb_writer = tb_writer
+        self.global_step = 0
+
+    def train(self, n_episodes: int, seed: int = 42):
+
+        obs, _ = self.env.reset(seed=seed)
+        prices = None
+        for ep in range(n_episodes):
+            prices = self.engine.compute_prices(prices, obs)
+            obs, reward, done, _, info = self.env.step(prices)
+            self.engine.update(obs, reward, done, info)
+        return self
+
+
+
+
+
+
+        return self.episode_metrics
+
+    def evaluate(self, n_episodes: int = 10, seed: int = 100) -> Dict:
+        """evaluate trained engine"""
+        results = {k: [] for k in ['total_reward', 'revenue_observed', 'revenue_oracle',
+                                   'agent_loss', 'ux_volatility', 'look_to_book']}
+        for ep in range(n_episodes):
+            metrics = self.run_episode(seed=seed + ep)
+            for k in results:                results[k].append(metrics[k])
+        return {k: (np.mean(v), np.std(v)) for k, v in results.items()}
+
+
+def make_env(fast: bool = True):
+    constraints = FastTrainingConstraints() if fast else BusinessLogicConstraints()
+    return PHANTOMEnv(constraints=constraints)
+
+
+def train_engine(engine_cls: Type[BasePricingEngine], env: PHANTOMEnv,
+                n_episodes: int, seed: int = 42,
+                tb_writer: Optional[SummaryWriter] = None) -> EngineTrainer:
+    constraints = env.constraints
+    engine = engine_cls(constraints=constraints, seed=seed)
+    trainer = EngineTrainer(engine, env, tb_writer=tb_writer)
+    trainer.train(n_episodes, seed=seed)
+    return trainer
+
+
+def save_trainer(trainer: EngineTrainer, path: Path):
+    """save engine state and metrics"""
+    path.parent.mkdir(parents=True, exist_ok=True)
+    with open(path, 'wb') as f:
+        pickle.dump({
+            'engine': trainer.engine,
+            'metrics': trainer.episode_metrics
+        }, f)
+    logger.info(f"Saved trainer to {path}")
+
+
+def load_trainer(path: Path, env: PHANTOMEnv,
+                 tb_writer: Optional[SummaryWriter] = None) -> EngineTrainer:
+    """load saved engine"""
+    with open(path, 'rb') as f:
+        data = pickle.load(f)
+    trainer = EngineTrainer(data['engine'], env, tb_writer=tb_writer)
+    trainer.episode_metrics = data['metrics']
+    return trainer
+
+
+if __name__ == "__main__":
+    base_dir = Path("./runs")
+    base_dir.mkdir(exist_ok=True)
+
+    engines = {
+        "Wild": WildPricingEngine,
+        "Static": StaticPricingEngine,
+#        "SimpleDemand": SimpleDemandEngine,
+        "RandomWalk": RandomWalkEngine,
+        "ThompsonSampling": ThompsonSamplingEngine,
+    }
+    defenses = [False, True]
+    n_train_episodes = 50
+    n_eval_episodes = 10
+    seed = 42
+    fast_mode = True
+
+    logger.info(f"Training config: {n_train_episodes} episodes per engine, fast_mode={fast_mode}")
+
+    trained_trainers = {}
+
+    for engine_name, engine_cls in engines.items():
+        for use_defense in defenses:
+            defense_label = "defense_on" if use_defense else "defense_off"
+            run_name = f"{engine_name}_{defense_label}"
+            log_dir = base_dir / run_name
+            log_dir.mkdir(parents=True, exist_ok=True)
+
+            logger.info(f"Training {engine_name} with defense={use_defense}")
+            logger.info(f"Log directory: {log_dir}")
+
+            env = make_env(fast=fast_mode)
+            tb_writer = SummaryWriter(log_dir=str(log_dir))
+            trainer = train_engine(engine_cls, env, n_train_episodes, seed, tb_writer=tb_writer)
+            tb_writer.close()
+
+            save_path = log_dir / "trainer.pkl"
+            save_trainer(trainer, save_path)
+
+            trained_trainers[run_name] = (trainer, env)
+
+    logger.info("Starting evaluation")
+
+    for run_name, (trainer, env) in trained_trainers.items():
+        logger.info(f"Evaluating {run_name}")
+        results = trainer.evaluate(n_episodes=n_eval_episodes, seed=seed + 1000)
+        for metric, (mean, std) in results.items():
+            logger.info(f"  {metric:20s}: {mean:10.2f} ± {std:6.2f}")
+
+    logger.info(f"Results saved to: {base_dir}")

tests/e2e/__init__.py

@@ -0,0 +1 @@
+"""E2E test suite for PHANTOM dynamic pricing pipeline."""

tests/e2e/fixtures.ts

@@ -0,0 +1,17 @@
+import { test as base } from '@playwright/test';
+
+type TestFixtures = {
+  backendUrl: string;
+  pricingUrl: string;
+};
+
+export const test = base.extend<TestFixtures>({
+  backendUrl: async ({}, use) => {
+    await use(process.env.BACKEND_URL || 'http://localhost:5000');
+  },
+  pricingUrl: async ({}, use) => {
+    await use(process.env.PRICING_PROVIDER_URL || 'http://localhost:5001');
+  },
+});
+
+export { expect } from '@playwright/test';

tests/e2e/helpers/api.ts

@@ -0,0 +1,69 @@
+interface PriceResponse {
+  price: number;
+  base_price: number;
+  markup: number;
+  model_version?: string;
+}
+
+export async function fetchPrice(
+  baseUrl: string,
+  productId: string,
+  mode: string = 'simple_surge',
+  sessionId?: string
+): Promise<PriceResponse> {
+  const params = new URLSearchParams();
+  if (sessionId) params.set('sessionId', sessionId);
+
+  const url = `${baseUrl}/api/pricing?mode=${mode}&productId=${productId}&${params}`;
+  const resp = await fetch(url);
+
+  if (!resp.ok) {
+    throw new Error(`Price fetch failed: ${resp.status}`);
+  }
+
+  return resp.json();
+}
+
+export async function waitForPriceChange(
+  baseUrl: string,
+  productId: string,
+  baselinePrice: number,
+  mode: string,
+  sessionId?: string,
+  maxRetries: number = 10,
+  pollInterval: number = 500
+): Promise<PriceResponse> {
+  for (let i = 0; i < maxRetries; i++) {
+    const priceResp = await fetchPrice(baseUrl, productId, mode, sessionId);
+    if (Math.abs(priceResp.price - baselinePrice) > 0.01) {
+      return priceResp;
+    }
+    await new Promise(r => setTimeout(r, pollInterval));
+  }
+
+  throw new Error(`Price did not change after ${maxRetries} retries`);
+}
+
+export async function ingestEvent(
+  baseUrl: string,
+  sessionId: string,
+  event: string,
+  productId?: string,
+  metadata?: Record<string, any>
+): Promise<void> {
+  const resp = await fetch(`${baseUrl}/api/ingest`, {
+    method: 'POST',
+    headers: { 'Content-Type': 'application/json' },
+    body: JSON.stringify({
+      sessionId,
+      event,
+      productId,
+      timestamp: new Date().toISOString(),
+      metadata,
+    }),
+  });
+
+  if (!resp.ok) {
+    throw new Error(`Event ingest failed: ${resp.status}`);
+  }
+}

tests/e2e/helpers/interactions.ts

@@ -0,0 +1,219 @@
+import { Page } from '@playwright/test';
+
+export async function getSessionId(page: Page): Promise<string | null> {
+  const cookies = await page.context().cookies();
+  const sessionCookie = cookies.find(c => c.name === 'phantom_session_id');
+  return sessionCookie?.value || null;
+}
+
+export async function verifySessionConsistency(page: Page, expectedSessionId: string): Promise<boolean> {
+  const currentSessionId = await getSessionId(page);
+  return currentSessionId === expectedSessionId;
+}
+
+export async function createFreshSession(page: Page, storeType: 'hotel' | 'airline' = 'hotel'): Promise<string> {
+  await page.context().clearCookies();
+  await page.goto('/');
+  await page.waitForLoadState('networkidle');
+  await page.waitForTimeout(500);
+
+  const sid = await getSessionId(page);
+  if (!sid) throw new Error('Session not created');
+  return sid;
+}
+
+interface SearchParams {
+  destination?: string;
+  checkIn?: string;
+  guests?: number;
+  rooms?: number;
+  origin?: string;
+  departure?: string;
+  adults?: number;
+}
+
+export async function performSearch(page: Page, params: SearchParams, storeType: 'hotel' | 'airline' = 'hotel' ): Promise<void> {
+  await page.waitForLoadState('networkidle');
+
+  if (storeType === 'hotel') {
+    const destInput = page.locator('input#destination');
+    await destInput.fill(params.destination || 'New York');
+
+    const checkInInput = page.locator('input#checkIn');
+    const checkInDate = params.checkIn || new Date(Date.now() + 7 * 86400000).toISOString().split('T')[0];
+    await checkInInput.fill(checkInDate);
+
+    const searchBtn = page.locator('button:has-text("Search Rooms")');
+    await searchBtn.click();
+  } else {
+    const originDropdown = page.locator('button:has-text("Select origin")').or(
+      page.locator('[id="origin"]').locator('button').first()
+    );
+    await originDropdown.click();
+    await page.waitForTimeout(200);
+    const originOption = page.locator(`button:has-text("${params.origin || 'JFK'}")`).first();
+    await originOption.click();
+    await page.waitForTimeout(200);
+
+    const destDropdown = page.locator('button:has-text("Select destination")').or(
+      page.locator('[id="destination"]').locator('button').first()
+    );
+    await destDropdown.click();
+    await page.waitForTimeout(200);
+    const destOption = page.locator(`button:has-text("${params.destination || 'LAX'}")`).first();
+    await destOption.click();
+    await page.waitForTimeout(200);
+
+    const departInput = page.locator('input#departDate');
+    const departDate = params.departure || new Date(Date.now() + 7 * 86400000).toISOString().split('T')[0];
+    await departInput.fill(departDate);
+
+    const searchBtn = page.locator('button:has-text("Search Flights")');
+    await searchBtn.click();
+  }
+
+  await page.waitForLoadState('networkidle');
+}
+
+export async function selectRandomProduct(page: Page, storeType: 'hotel' | 'airline' = 'hotel'): Promise<string> {
+  await page.waitForLoadState('networkidle');
+
+  const cardClass = storeType === 'hotel' ? '.hotel-card' : '.flight-card';
+  const productCards = page.locator(cardClass);
+
+  const count = await productCards.count();
+  if (count === 0) throw new Error('No products found on listing page');
+
+  const randomIdx = Math.floor(Math.random() * count);
+  return randomIdx.toString();
+}
+
+export async function openProductFromListing(page: Page, productId?: string): Promise<string> {
+  await page.waitForLoadState('networkidle');
+
+  const hotelCards = page.locator('.hotel-card');
+  const flightCards = page.locator('.flight-card');
+
+  const hotelCount = await hotelCards.count();
+  const flightCount = await flightCards.count();
+
+  let productCards;
+  if (hotelCount > 0) {
+    productCards = hotelCards;
+  } else if (flightCount > 0) {
+    productCards = flightCards;
+  } else {
+    throw new Error('No products found on listing page');
+  }
+
+  const count = await productCards.count();
+  const randomIdx = productId ? 0 : Math.floor(Math.random() * count);
+  await productCards.nth(randomIdx).click();
+
+  await page.waitForLoadState('networkidle');
+
+  const url = page.url();
+  const match = url.match(/\/products\/([^/?]+)/);
+  if (!match) throw new Error('Cannot parse product ID from URL after navigation');
+
+  return match[1];
+}
+
+export async function getPriceFromDOM(page: Page): Promise<number> {
+  await page.waitForLoadState('networkidle');
+
+  await page.waitForSelector('.price-amount', { timeout: 15000 }).catch(() => null);
+
+  const priceSelectors = [
+    '.price-amount',
+    '.price-display',
+    '[data-testid="price"]',
+    '[data-price]',
+  ];
+
+  for (const selector of priceSelectors) {
+    const priceEl = page.locator(selector).first();
+    if (await priceEl.count() > 0) {
+      const text = await priceEl.textContent();
+      if (!text) continue;
+
+      const match = text.match(/[\$]?\s*([\d,]+(?:\.\d{2})?)/);
+      if (match) {
+        const priceStr = match[1].replace(/,/g, '');
+        return parseFloat(priceStr);
+      }
+    }
+  }
+
+  const dataPrice = await page.locator('[data-price]').first().getAttribute('data-price').catch(() => null);
+  if (dataPrice) return parseFloat(dataPrice);
+
+  throw new Error('Cannot extract price from DOM');
+}
+
+export async function navigateToProduct(page: Page,productId: string,storeType: 'hotel' | 'airline' = 'hotel'): Promise<void> {
+  await page.goto(`/products/${productId}`);
+  await page.waitForLoadState('networkidle');
+}
+
+export async function viewProductViaFlow(page: Page, storeType: 'hotel' | 'airline' = 'hotel', searchParams?: SearchParams): Promise<string> {
+  const params = new URLSearchParams();
+  params.set('dateIndex', '7');
+
+  if (storeType === 'hotel') {
+    params.set('destination', searchParams?.destination || 'New York');
+    params.set('adults', '2');
+    params.set('rooms', '1');
+  } else {
+    params.set('origin', searchParams?.origin || 'JFK');
+    params.set('destination', searchParams?.destination || 'LAX');
+    params.set('adults', '1');
+    params.set('children', '0');
+    params.set('infants', '0');
+  }
+
+  await page.goto(`/products?${params.toString()}`);
+  await page.waitForLoadState('networkidle');
+
+  const productId = await openProductFromListing(page);
+  await page.waitForTimeout(500);
+  return productId;
+}
+
+export async function rapidViewProductViaFlow(page: Page, count: number, delayMs: number = 100, storeType: 'hotel' | 'airline' = 'hotel'): Promise<string[]> {
+  const productIds: string[] = [];
+
+  for (let i = 0; i < count; i++) {
+    const productId = await viewProductViaFlow(page, storeType);
+    productIds.push(productId);
+
+    await page.waitForTimeout(delayMs);
+  }
+
+  return productIds;
+}
+
+export async function humanLikeViewProduct(page: Page, storeType: 'hotel' | 'airline' = 'hotel'
+): Promise<string> {
+  const productId = await viewProductViaFlow(page, storeType);
+
+  await page.hover('h1');
+  await page.waitForTimeout(800 + Math.random() * 400);
+
+  await page.mouse.wheel(0, 200);
+  await page.waitForTimeout(500 + Math.random() * 300);
+
+  const paragraphs = await page.locator('p').all();
+  if (paragraphs.length > 0) {
+    await paragraphs[0].hover();
+    await page.waitForTimeout(600 + Math.random() * 400);
+  }
+
+  return productId;
+}
+
+export async function addToCart(page: Page): Promise<void> {
+  const addBtn = page.locator('button:has-text("Add to Cart")');
+  await addBtn.click();
+  await page.waitForTimeout(500);
+}

tests/e2e/helpers/kafka.ts

@@ -0,0 +1,39 @@
+interface InteractionEvent {
+  sessionId: string;
+  event: string;
+  productId?: string;
+  timestamp: string;
+  metadata?: Record<string, any>;
+}
+
+const dumpKafkaTopic = async (backendUrl: string, topic: string) => {
+  const resp = await fetch(`${backendUrl}/api/kafka/dump?topic=${topic}`);
+  if (!resp.ok) throw new Error(`Kafka dump failed: ${resp.status}`);
+  const { data = [] } = await resp.json();
+  return data as any[];
+};
+
+export const waitForInteractionEvent = async (
+  backendUrl: string,
+  sessionId: string,
+  eventType: string,
+  maxRetries = 10,
+  pollInterval = 500
+): Promise<InteractionEvent | null> => {
+  for (let i = 0; i < maxRetries; i++) {
+    const msgs = await dumpKafkaTopic(backendUrl, "user-interactions");
+    const hit = msgs.find(m => m.sessionId === sessionId && m.event === eventType);
+    if (hit) return hit as InteractionEvent;
+    await new Promise<void>(r => setTimeout(r, pollInterval));
+  }
+  return null;
+};
+
+export const countProductViews = async (backendUrl: string, productId: string) =>
+  (await dumpKafkaTopic(backendUrl, "user-interactions")).reduce(
+    (n, m) => n + (m.productId === productId && m.event === "view_item_page" ? 1 : 0),
+    0
+  );
+
+export const getSessionEvents = async (backendUrl: string, sessionId: string) =>
+  (await dumpKafkaTopic(backendUrl, "user-interactions")).filter(m => m.sessionId === sessionId);

tests/e2e/package.json

@@ -0,0 +1,19 @@
+{
+  "name": "e2e",
+  "version": "1.0.0",
+  "main": "index.js",
+  "scripts": {
+    "test": "playwright test",
+    "test:ui": "playwright test --ui",
+    "test:debug": "playwright test --debug"
+  },
+  "keywords": [],
+  "author": "",
+  "license": "ISC",
+  "description": "",
+  "devDependencies": {
+    "@playwright/test": "^1.57.0",
+    "@types/node": "^25.0.6",
+    "typescript": "^5.9.3"
+  }
+}

tests/e2e/playwright.config.ts

@@ -0,0 +1,25 @@
+import { defineConfig, devices } from '@playwright/test';
+
+export default defineConfig({
+  testDir: './scenarios',
+  fullyParallel: true,
+  forbidOnly: !!process.env.CI,
+  retries: 0,
+  workers: 1,
+  reporter: 'list',
+  use: {
+    baseURL: process.env.WEB_URL || 'http://localhost:3000',
+    trace: 'retain-on-failure',
+    screenshot: 'only-on-failure',
+  },
+  timeout: 180000,
+  expect: {
+    timeout: 10000,
+  },
+  projects: [
+    {
+      name: 'chromium',
+      use: { ...devices['Desktop Chrome'] },
+    },
+  ],
+});

tests/e2e/scenarios/session-aware.spec.ts

@@ -0,0 +1,163 @@
+import { test, expect } from '../fixtures';
+import {
+  createFreshSession,
+  viewProductViaFlow,
+  rapidViewProductViaFlow,
+  humanLikeViewProduct,
+  getPriceFromDOM,
+  verifySessionConsistency,
+  addToCart,
+} from '../helpers/interactions';
+import { getSessionEvents } from '../helpers/kafka';
+import { runSessionPricing } from '../helpers/airflow';
+
+test.describe('SessionAwarePricer E2E', () => {
+  const STORE_TYPE = 'hotel';
+
+  test('baseline: human-like behavior maintains base price', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId1 = await humanLikeViewProduct(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+
+    await page.waitForTimeout(1500);
+
+    const productId2 = await humanLikeViewProduct(page, STORE_TYPE);
+
+    await runSessionPricing(STORE_TYPE);
+
+    const secondPrice = await getPriceFromDOM(page);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+
+    expect(Math.abs(secondPrice - baselinePrice) / baselinePrice).toBeLessThan(0.1);
+  });
+
+  test('agent detection: rapid robot-like behavior increases price', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await page.waitForTimeout(500);
+
+    await rapidViewProductViaFlow(page, 8, 100, STORE_TYPE);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+
+    await page.waitForTimeout(1000);
+
+    const events = await getSessionEvents(backendUrl, sessionId);
+    expect(events.length).toBeGreaterThanOrEqual(8);
+
+    await runSessionPricing(STORE_TYPE);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const agentPrice = await getPriceFromDOM(page);
+
+    expect(agentPrice).toBeGreaterThan(baselinePrice);
+    expect((agentPrice - baselinePrice) / baselinePrice).toBeGreaterThan(0.01);
+  });
+
+  test('velocity threshold: high event rate triggers detection', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 10, 80, STORE_TYPE);
+
+    const events = await getSessionEvents(backendUrl, sessionId);
+    expect(events.length).toBeGreaterThanOrEqual(10);
+
+    await runSessionPricing(STORE_TYPE);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const agentPrice = await getPriceFromDOM(page);
+
+    expect(agentPrice).toBeGreaterThan(baselinePrice);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('cart ratio: high cart/view ratio signals intent', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await page.waitForTimeout(500);
+    await addToCart(page);
+
+    await page.waitForTimeout(2000);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const cartPrice = await getPriceFromDOM(page);
+
+    expect(cartPrice).toBeGreaterThanOrEqual(baselinePrice);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('mixed behavior: occasional fast actions tolerated', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId1 = await humanLikeViewProduct(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await page.waitForTimeout(1200);
+
+    await rapidViewProductViaFlow(page, 2, 150, STORE_TYPE);
+
+    await page.waitForTimeout(1000);
+    await humanLikeViewProduct(page, STORE_TYPE);
+
+    await runSessionPricing(STORE_TYPE);
+
+    const finalPrice = await getPriceFromDOM(page);
+
+    expect(Math.abs(finalPrice - baselinePrice) / baselinePrice).toBeLessThan(0.3);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('session isolation: agent behavior in one session does not affect others', async ({
+    page,
+    context,
+    backendUrl,
+  }) => {
+    const sessionIdA = await createFreshSession(page, STORE_TYPE);
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const basePrice = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 10, 100, STORE_TYPE);
+    await page.waitForTimeout(2000);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const agentPrice = await getPriceFromDOM(page);
+    expect(agentPrice).toBeGreaterThan(basePrice * 0.99);
+
+    const page2 = await context.newPage();
+    const sessionIdB = await createFreshSession(page2, STORE_TYPE);
+
+    await page2.goto(`/products/${productId}`);
+    await page2.waitForLoadState('networkidle');
+    const cleanPrice = await getPriceFromDOM(page2);
+
+    expect(Math.abs(cleanPrice - basePrice) / basePrice).toBeLessThan(0.1);
+    expect(sessionIdA).not.toBe(sessionIdB);
+  });
+
+  test('session persistence: session ID maintained across views', async ({ page }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    await viewProductViaFlow(page, STORE_TYPE);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+
+    await viewProductViaFlow(page, STORE_TYPE);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+
+    await viewProductViaFlow(page, STORE_TYPE);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+});

tests/e2e/scenarios/surge-pricing.spec.ts

@@ -0,0 +1,118 @@
+import { test, expect } from '../fixtures';
+import {
+  createFreshSession,
+  viewProductViaFlow,
+  rapidViewProductViaFlow,
+  getPriceFromDOM,
+  verifySessionConsistency,
+} from '../helpers/interactions';
+import { waitForInteractionEvent, countProductViews } from '../helpers/kafka';
+import { runSurgePricing } from '../helpers/airflow';
+
+test.describe('SimpleSurgePricer E2E', () => {
+  const STORE_TYPE = 'hotel';
+
+  test('baseline: initial price equals base price', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const price = await getPriceFromDOM(page);
+
+    expect(price).toBeGreaterThan(0);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('surge: rapid views trigger price increase', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 5, 200, STORE_TYPE);
+
+    await page.waitForTimeout(1000);
+
+    const evt = await waitForInteractionEvent(backendUrl, sessionId, 'view_item_page');
+    expect(evt).not.toBeNull();
+
+    const viewCount = await countProductViews(backendUrl, productId);
+    expect(viewCount).toBeGreaterThanOrEqual(5);
+
+    await runSurgePricing(STORE_TYPE, 3, 1);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const surgedPrice = await getPriceFromDOM(page);
+
+    expect(surgedPrice).toBeGreaterThan(baselinePrice);
+    expect((surgedPrice - baselinePrice) / baselinePrice).toBeGreaterThan(0.01);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('threshold: price unchanged below threshold', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 2, 300, STORE_TYPE);
+
+    await page.waitForTimeout(1500);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const currentPrice = await getPriceFromDOM(page);
+
+    expect(Math.abs(currentPrice - baselinePrice) / baselinePrice).toBeLessThan(0.05);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('window: surge decays after window expires', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productId = await viewProductViaFlow(page, STORE_TYPE);
+    const baselinePrice = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 5, 150, STORE_TYPE);
+
+    await page.waitForTimeout(1000);
+
+    await runSurgePricing(STORE_TYPE, 3, 1);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const surgedPrice = await getPriceFromDOM(page);
+    expect(surgedPrice).toBeGreaterThan(baselinePrice);
+
+    await page.waitForTimeout(12000);
+
+    await runSurgePricing(STORE_TYPE, 3, 1);
+
+    await page.goto(`/products/${productId}`);
+    await page.waitForLoadState('networkidle');
+    const decayedPrice = await getPriceFromDOM(page);
+    expect(decayedPrice).toBeLessThan(surgedPrice);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+
+  test('isolation: different products have independent surge', async ({ page, backendUrl }) => {
+    const sessionId = await createFreshSession(page, STORE_TYPE);
+
+    const productIdA = await viewProductViaFlow(page, STORE_TYPE);
+    const basePriceA = await getPriceFromDOM(page);
+
+    await rapidViewProductViaFlow(page, 5, 200, STORE_TYPE);
+    await page.waitForTimeout(2000);
+
+    await page.goto(`/products/${productIdA}`);
+    await page.waitForLoadState('networkidle');
+    const surgedPriceA = await getPriceFromDOM(page);
+
+    const productIdB = await viewProductViaFlow(page, STORE_TYPE);
+    const priceB = await getPriceFromDOM(page);
+
+    expect(surgedPriceA).toBeGreaterThan(basePriceA * 0.99);
+    expect(productIdA).not.toBe(productIdB);
+    expect(await verifySessionConsistency(page, sessionId)).toBeTruthy();
+  });
+});

tests/e2e/tsconfig.json

@@ -0,0 +1,15 @@
+{
+  "compilerOptions": {
+    "target": "ES2022",
+    "module": "commonjs",
+    "lib": ["ES2022"],
+    "strict": true,
+    "esModuleInterop": true,
+    "skipLibCheck": true,
+    "forceConsistentCasingInFileNames": true,
+    "resolveJsonModule": true,
+    "types": ["node", "@playwright/test"]
+  },
+  "include": ["**/*.ts"],
+  "exclude": ["node_modules"]
+}

web/package-lock.json

@@ -10,7 +10,7 @@
       "dependencies": {
         "@supabase/ssr": "^0.7.0",
         "@supabase/supabase-js": "^2.81.1",
-        "next": "16.0.0",
+        "next": "^16.0.0",
         "react": "19.2.0",
         "react-dom": "19.2.0",
         "zod": "^4.1.12"
@@ -526,15 +526,15 @@
       }
     },
     "node_modules/@next/env": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/env/-/env-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/env/-/env-16.0.7.tgz",
-      "integrity": "sha512-s5j2iFGp38QsG1LWRQaE2iUY3h1jc014/melHFfLdrsMJPqxqDQwWNwyQTcNoUSGZlCVZuM7t7JDMmSyRilsnA==",
+      "integrity": "sha512-gpaNgUh5nftFKRkRQGnVi5dpcYSKGcZZkQffZ172OrG/XkrnS7UBTQ648YY+8ME92cC4IojpI2LqTC8sTDhAaw==",
       "license": "MIT"
     },
     "node_modules/@next/swc-darwin-arm64": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-darwin-arm64/-/swc-darwin-arm64-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-darwin-arm64/-/swc-darwin-arm64-16.0.7.tgz",
-      "integrity": "sha512-/CntqDCnk5w2qIwMiF0a9r6+9qunZzFmU0cBX4T82LOflE72zzH6gnOjCwUXYKOBlQi8OpP/rMj8cBIr18x4TA==",
+      "integrity": "sha512-LlDtCYOEj/rfSnEn/Idi+j1QKHxY9BJFmxx7108A6D8K0SB+bNgfYQATPk/4LqOl4C0Wo3LACg2ie6s7xqMpJg==",
       "cpu": [
         "arm64"
       ],
@@ -548,9 +548,9 @@
       }
     },
     "node_modules/@next/swc-darwin-x64": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-darwin-x64/-/swc-darwin-x64-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-darwin-x64/-/swc-darwin-x64-16.0.7.tgz",
-      "integrity": "sha512-hB4GZnJGKa8m4efvTGNyii6qs76vTNl+3dKHTCAUaksN6KjYy4iEO3Q5ira405NW2PKb3EcqWiRaL9DrYJfMHg==",
+      "integrity": "sha512-rtZ7BhnVvO1ICf3QzfW9H3aPz7GhBrnSIMZyr4Qy6boXF0b5E3QLs+cvJmg3PsTCG2M1PBoC+DANUi4wCOKXpA==",
       "cpu": [
         "x64"
       ],
@@ -564,9 +564,9 @@
       }
     },
     "node_modules/@next/swc-linux-arm64-gnu": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-gnu/-/swc-linux-arm64-gnu-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-gnu/-/swc-linux-arm64-gnu-16.0.7.tgz",
-      "integrity": "sha512-E2IHMdE+C1k+nUgndM13/BY/iJY9KGCphCftMh7SXWcaQqExq/pJU/1Hgn8n/tFwSoLoYC/yUghOv97tAsIxqg==",
+      "integrity": "sha512-mloD5WcPIeIeeZqAIP5c2kdaTa6StwP4/2EGy1mUw8HiexSHGK/jcM7lFuS3u3i2zn+xH9+wXJs6njO7VrAqww==",
       "cpu": [
         "arm64"
       ],
@@ -580,9 +580,9 @@
       }
     },
     "node_modules/@next/swc-linux-arm64-musl": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-musl/-/swc-linux-arm64-musl-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-musl/-/swc-linux-arm64-musl-16.0.7.tgz",
-      "integrity": "sha512-xzgl7c7BVk4+7PDWldU+On2nlwnGgFqJ1siWp3/8S0KBBLCjonB6zwJYPtl4MUY7YZJrzzumdUpUoquu5zk8vg==",
+      "integrity": "sha512-+ksWNrZrthisXuo9gd1XnjHRowCbMtl/YgMpbRvFeDEqEBd523YHPWpBuDjomod88U8Xliw5DHhekBC3EOOd9g==",
       "cpu": [
         "arm64"
       ],
@@ -596,9 +596,9 @@
       }
     },
     "node_modules/@next/swc-linux-x64-gnu": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-gnu/-/swc-linux-x64-gnu-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-gnu/-/swc-linux-x64-gnu-16.0.7.tgz",
-      "integrity": "sha512-sdyOg4cbiCw7YUr0F/7ya42oiVBXLD21EYkSwN+PhE4csJH4MSXUsYyslliiiBwkM+KsuQH/y9wuxVz6s7Nstg==",
+      "integrity": "sha512-4WtJU5cRDxpEE44Ana2Xro1284hnyVpBb62lIpU5k85D8xXxatT+rXxBgPkc7C1XwkZMWpK5rXLXTh9PFipWsA==",
       "cpu": [
         "x64"
       ],
@@ -612,9 +612,9 @@
       }
     },
     "node_modules/@next/swc-linux-x64-musl": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-musl/-/swc-linux-x64-musl-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-musl/-/swc-linux-x64-musl-16.0.7.tgz",
-      "integrity": "sha512-IAXv3OBYqVaNOgyd3kxR4L3msuhmSy1bcchPHxDOjypG33i2yDWvGBwFD94OuuTjjTt/7cuIKtAmoOOml6kfbg==",
+      "integrity": "sha512-HYlhqIP6kBPXalW2dbMTSuB4+8fe+j9juyxwfMwCe9kQPPeiyFn7NMjNfoFOfJ2eXkeQsoUGXg+O2SE3m4Qg2w==",
       "cpu": [
         "x64"
       ],
@@ -628,9 +628,9 @@
       }
     },
     "node_modules/@next/swc-win32-arm64-msvc": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-win32-arm64-msvc/-/swc-win32-arm64-msvc-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-win32-arm64-msvc/-/swc-win32-arm64-msvc-16.0.7.tgz",
-      "integrity": "sha512-bmo3ncIJKUS9PWK1JD9pEVv0yuvp1KPuOsyJTHXTv8KDrEmgV/K+U0C75rl9rhIaODcS7JEb6/7eJhdwXI0XmA==",
+      "integrity": "sha512-EviG+43iOoBRZg9deGauXExjRphhuYmIOJ12b9sAPy0eQ6iwcPxfED2asb/s2/yiLYOdm37kPaiZu8uXSYPs0Q==",
       "cpu": [
         "arm64"
       ],
@@ -644,9 +644,9 @@
       }
     },
     "node_modules/@next/swc-win32-x64-msvc": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-win32-x64-msvc/-/swc-win32-x64-msvc-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/@next/swc-win32-x64-msvc/-/swc-win32-x64-msvc-16.0.7.tgz",
-      "integrity": "sha512-O1cJbT+lZp+cTjYyZGiDwsOjO3UHHzSqobkPNipdlnnuPb1swfcuY6r3p8dsKU4hAIEO4cO67ZCfVVH/M1ETXA==",
+      "integrity": "sha512-gniPjy55zp5Eg0896qSrf3yB1dw4F/3s8VK1ephdsZZ129j2n6e1WqCbE2YgcKhW9hPB9TVZENugquWJD5x0ug==",
       "cpu": [
         "x64"
       ],
@@ -1447,12 +1447,12 @@
       }
     },
     "node_modules/next": {
-      "version": "16.0.0",
+      "version": "16.0.7",
-      "resolved": "https://registry.npmjs.org/next/-/next-16.0.0.tgz",
+      "resolved": "https://registry.npmjs.org/next/-/next-16.0.7.tgz",
-      "integrity": "sha512-nYohiNdxGu4OmBzggxy9rczmjIGI+TpR5vbKTsE1HqYwNm1B+YSiugSrFguX6omMOKnDHAmBPY4+8TNJk0Idyg==",
+      "integrity": "sha512-3mBRJyPxT4LOxAJI6IsXeFtKfiJUbjCLgvXO02fV8Wy/lIhPvP94Fe7dGhUgHXcQy4sSuYwQNcOLhIfOm0rL0A==",
       "license": "MIT",
       "dependencies": {
-        "@next/env": "16.0.0",
+        "@next/env": "16.0.7",
         "@swc/helpers": "0.5.15",
         "caniuse-lite": "^1.0.30001579",
         "postcss": "8.4.31",
@@ -1465,14 +1465,14 @@
         "node": ">=20.9.0"
       },
       "optionalDependencies": {
-        "@next/swc-darwin-arm64": "16.0.0",
+        "@next/swc-darwin-arm64": "16.0.7",
-        "@next/swc-darwin-x64": "16.0.0",
+        "@next/swc-darwin-x64": "16.0.7",
-        "@next/swc-linux-arm64-gnu": "16.0.0",
+        "@next/swc-linux-arm64-gnu": "16.0.7",
-        "@next/swc-linux-arm64-musl": "16.0.0",
+        "@next/swc-linux-arm64-musl": "16.0.7",
-        "@next/swc-linux-x64-gnu": "16.0.0",
+        "@next/swc-linux-x64-gnu": "16.0.7",
-        "@next/swc-linux-x64-musl": "16.0.0",
+        "@next/swc-linux-x64-musl": "16.0.7",
-        "@next/swc-win32-arm64-msvc": "16.0.0",
+        "@next/swc-win32-arm64-msvc": "16.0.7",
-        "@next/swc-win32-x64-msvc": "16.0.0",
+        "@next/swc-win32-x64-msvc": "16.0.7",
         "sharp": "^0.34.4"
       },
       "peerDependencies": {

web/package.json

@@ -10,7 +10,7 @@
   "dependencies": {
     "@supabase/ssr": "^0.7.0",
     "@supabase/supabase-js": "^2.81.1",
-    "next": "16.0.0",
+    "next": "^16.0.0",
     "react": "19.2.0",
     "react-dom": "19.2.0",
     "zod": "^4.1.12"

web/src/app/airline/checkout/page.tsx

@@ -0,0 +1,11 @@
+export default function AirlineCheckout() {
+  return (
+    <div className="min-h-screen flex items-center justify-center bg-gradient-to-br from-sky-50 to-blue-50">
+      <div className="text-center p-8">
+        <h1 className="text-4xl font-light text-gray-800 mb-4">
+          Thank you for flying with us
+        </h1>
+      </div>
+    </div>
+  );
+}

web/src/app/api/ingest/route.ts

@@ -7,7 +7,7 @@ export async function POST(req: NextRequest) {
   try {
     const body = await req.json();
 
-    const storeMode = process.env.STORE_MODE || 'hotel';
+    const storeMode = process.env.NEXT_PUBLIC_STORE_MODE || process.env.STORE_MODE || 'hotel';
     const userAgent = req.headers.get('user-agent') || undefined;
 
     const event: EventBase = {

web/src/app/api/pricing/route.ts

@@ -11,7 +11,7 @@ export async function GET(req: NextRequest) {
     const productId = searchParams.get('productId');
     const sessionId = searchParams.get('sessionId');
     const experimentId = searchParams.get('experimentId');
-    const storeMode = process.env.NEXT_PUBLIC_STORE_MODE || 'shop';
+    const storeMode = process.env.NEXT_PUBLIC_STORE_MODE || process.env.STORE_MODE || 'hotel';
 
     if (!productId) {
         return NextResponse.json(
@@ -30,6 +30,8 @@ export async function GET(req: NextRequest) {
     const providerUrl = process.env.PRICING_PROVIDER_URL || 'http://localhost:5001';
     try {
         const queryParams = new URLSearchParams();
+        // THIS is our entry point into the dynamic pricing where we reference the context of the sesion and experiment and ask for a price to assign to the trajectory which is expressed
+        // The whole pipeline gets triggered from here.
         if (sessionId) queryParams.append('sessionId', sessionId);
         if (experimentId) queryParams.append('experimentId', experimentId);
 
@@ -55,11 +57,11 @@ export async function GET(req: NextRequest) {
         price = Math.round(randomBase * 100) / 100;
     }
 
-    // log price to kafka for elasticity computation
+    // log price to kafka asynchronously (non-blocking)
     if (sessionId) {
         const backendUrl = process.env.BACKEND_URL || 'http://localhost:5000';
-        try {
+        // fire and forget - don't await to avoid blocking response
-            await fetch(`${backendUrl}/api/kafka/price-log`, {
+        fetch(`${backendUrl}/api/kafka/price-log`, {
             method: 'POST',
             headers: { 'Content-Type': 'application/json' },
             body: JSON.stringify({
@@ -70,10 +72,11 @@ export async function GET(req: NextRequest) {
                 storeMode,
                 ts: timestamp,
             }),
-            });
+        }).catch(err => {
-        } catch (err) {
+            if (process.env.NODE_ENV === 'development') {
                 console.error('[price-log-error]', err);
             }
+        });
     }
 
     if (process.env.NODE_ENV === 'development') {

web/src/app/cart/page.tsx

@@ -32,7 +32,8 @@ export default function CartPage() {
                     {itemCount > 0 && (
                         <button
                             onClick={clearCart}
-                            className="text-sm text-red-600 hover:underline"
+                            className="text-sm hover:underline"
+                            style={{ color: 'var(--accent-warning)' }}
                         >
                             Clear cart
                         </button>
@@ -42,7 +43,7 @@ export default function CartPage() {
                 {itemCount === 0 ? (
                     <div className="text-center py-12">
                         <p className="text-gray-500 mb-4">Your cart is empty</p>
-                        <a href="/" className="text-blue-600 hover:underline">Browse our selection</a>
+                        <a href="/" className="hover:underline" style={{ color: 'var(--text-accent)' }}>Browse our selection</a>
                     </div>
                 ) : (
                     <>
@@ -54,15 +55,11 @@ export default function CartPage() {
                                 >
                                     <div className="flex-1">
                                         <div className="flex items-center gap-2 mb-1">
-                                            <span className="px-2 py-0.5 text-xs font-medium rounded bg-blue-100 text-blue-800">
-                                                {item.type}
-                                            </span>
                                             <h3 className="font-semibold">{item.name}</h3>
                                         </div>
 
                                         {item.type === 'hotel' && (
                                             <div className="text-sm text-gray-600">
-                                                <p>{String(item.metadata.roomType)}</p>
                                                 <p>{String(item.metadata.checkIn)} - {String(item.metadata.checkOut)}</p>
                                                 <p>{String(item.metadata.nights)} night{Number(item.metadata.nights) > 1 ? 's' : ''}</p>
                                             </div>
@@ -81,7 +78,8 @@ export default function CartPage() {
                                         <p className="text-xl font-bold mb-2">${item.price}</p>
                                         <button
                                             onClick={() => handleRemove(item.id, item.type)}
-                                            className="text-sm text-red-600 hover:underline"
+                                            className="text-sm hover:underline"
+                                            style={{ color: 'var(--accent-warning)' }}
                                         >
                                             Remove
                                         </button>
@@ -96,8 +94,11 @@ export default function CartPage() {
                                 <span className="text-3xl font-bold">${total.toFixed(2)}</span>
                             </div>
                             <button
-                                onClick={() => dispatchInteraction('checkout_start', undefined, { total, itemCount })}
+                                onClick={() => {
-                                className="w-full py-3 bg-blue-600 hover:bg-blue-700 text-white rounded-lg font-medium transition-colors"
+                                    dispatchInteraction('checkout_start', undefined, { total, itemCount });
+                                    window.location.href = '/checkout';
+                                }}
+                                className="btn-primary w-full"
                             >
                                 Proceed to Checkout
                             </button>

web/src/app/globals.css

@@ -8,6 +8,9 @@
   --bg-secondary: #f5f5f5;
   --text-primary: #333333;
   --text-secondary: #666666;
+  --accent-primary: #007aff;
+  --accent-primary-hover: #0051d5;
+  --accent-primary-light: #e6f2ff;
   --spacing-sm: 8px;
   --spacing-md: 16px;
   --spacing-lg: 32px;

web/src/app/hotel/checkout/page.tsx

@@ -0,0 +1,11 @@
+export default function HotelCheckout() {
+  return (
+    <div className="min-h-screen flex items-center justify-center bg-gradient-to-br from-blue-50 to-indigo-50">
+      <div className="text-center p-8">
+        <h1 className="text-4xl font-light text-gray-800 mb-4">
+          Thank you for staying with us
+        </h1>
+      </div>
+    </div>
+  );
+}

web/src/app/layout.tsx

@@ -15,8 +15,8 @@ const geistMono = Geist_Mono({
 });
 
 export const metadata: Metadata = {
-  title: "Create Next App",
+  title: "Travel Booking Platform",
-  description: "Generated by create next app",
+  description: "Book flights and hotels with dynamic pricing",
 };
 
 export default function RootLayout({

web/src/components/feats/airline/AirlineHero.tsx

@@ -2,10 +2,20 @@
 
 import { useState, FormEvent } from 'react';
 import { useRouter } from 'next/navigation';
-import { Button, Label, Input, DateInput, RadioGroup, Dropdown, DropdownCounter } from '@/components/ui';
+import { Button, Label, DateInput, Dropdown, DropdownCounter, SelectDropdown, SelectOption } from '@/components/ui';
 import { dateToDaysFromToday } from '@/lib/airline-utils';
 
-type TripType = 'roundtrip' | 'oneway' | 'multicity';
+const CITIES: SelectOption[] = [
+  { value: 'JFK', label: 'New York (JFK)', sublabel: 'John F. Kennedy International' },
+  { value: 'LAX', label: 'Los Angeles (LAX)', sublabel: 'Los Angeles International' },
+  { value: 'ORD', label: 'Chicago (ORD)', sublabel: "O'Hare International" },
+  { value: 'MIA', label: 'Miami (MIA)', sublabel: 'Miami International' },
+  { value: 'SFO', label: 'San Francisco (SFO)', sublabel: 'San Francisco International' },
+  { value: 'SEA', label: 'Seattle (SEA)', sublabel: 'Seattle-Tacoma International' },
+  { value: 'ATL', label: 'Atlanta (ATL)', sublabel: 'Hartsfield-Jackson International' },
+  { value: 'DFW', label: 'Dallas (DFW)', sublabel: 'Dallas/Fort Worth International' },
+];
+
 
 const PlaneIcon = () => (
   <svg className="w-5 h-5" fill="none" stroke="currentColor" viewBox="0 0 24 24">
@@ -22,11 +32,9 @@ const LocationIcon = () => (
 
 export default function AirlineHero() {
   const router = useRouter();
-  const [tripType, setTripType] = useState<TripType>('roundtrip');
   const [origin, setOrigin] = useState('');
   const [destination, setDestination] = useState('');
   const [departDate, setDepartDate] = useState('');
-  const [returnDate, setReturnDate] = useState('');
   const [passengers, setPassengers] = useState({ adults: 1, children: 0, infants: 0 });
 
   const handleSearch = (e: FormEvent) => {
@@ -40,8 +48,6 @@ export default function AirlineHero() {
 
     if (origin) params.set('origin', origin);
     if (destination) params.set('destination', destination);
-    if (tripType !== 'roundtrip') params.set('tripType', tripType);
-    if (returnDate && tripType === 'roundtrip') params.set('returnDate', returnDate);
 
     params.set('adults', passengers.adults.toString());
     params.set('children', passengers.children.toString());
@@ -66,28 +72,15 @@ export default function AirlineHero() {
 
         <div className="search-form">
           <form onSubmit={handleSearch}>
-            <div className="mb-6">
+            <div className="grid grid-cols-1 sm:grid-cols-2 lg:grid-cols-3 gap-4">
-              <RadioGroup
-                name="tripType"
-                value={tripType}
-                onChange={setTripType}
-                options={[
-                  { value: 'roundtrip', label: 'Round-trip' },
-                  { value: 'oneway', label: 'One-way' },
-                  { value: 'multicity', label: 'Multi-city' },
-                ]}
-              />
-            </div>
-
-            <div className="grid grid-cols-1 sm:grid-cols-2 lg:grid-cols-4 gap-4">
               <div>
                 <Label htmlFor="origin">From</Label>
-                <Input
+                <SelectDropdown
-                  type="text"
                   id="origin"
                   value={origin}
-                  onChange={(e) => setOrigin(e.target.value)}
+                  onChange={setOrigin}
-                  placeholder="Airport or city"
+                  options={CITIES}
+                  placeholder="Select origin"
                   icon={<PlaneIcon />}
                   required
                 />
@@ -95,12 +88,12 @@ export default function AirlineHero() {
 
               <div>
                 <Label htmlFor="destination">To</Label>
-                <Input
+                <SelectDropdown
-                  type="text"
                   id="destination"
                   value={destination}
-                  onChange={(e) => setDestination(e.target.value)}
+                  onChange={setDestination}
-                  placeholder="Airport or city"
+                  options={CITIES}
+                  placeholder="Select destination"
                   icon={<LocationIcon />}
                   required
                 />
@@ -115,20 +108,6 @@ export default function AirlineHero() {
                   required
                 />
               </div>
-
-              <div>
-                <Label htmlFor="returnDate">Return</Label>
-                {tripType === 'roundtrip' ? (
-                  <DateInput
-                    id="returnDate"
-                    value={returnDate}
-                    onChange={(e) => setReturnDate(e.target.value)}
-                    required
-                  />
-                ) : (
-                  <DateInput id="returnDate" disabled />
-                )}
-              </div>
             </div>
 
             <div className="grid grid-cols-4 sm:grid-cols-3 lg:grid-cols-4 gap-4 mt-4">

web/src/components/feats/hotel/HotelCard.tsx

@@ -2,6 +2,7 @@
 
 import type { EventName } from '@/lib/events';
 import type { Hotel } from '@/lib/hotel-utils';
+import { getHotelImageUrl } from '@/lib/hotel-utils';
 import { useHoverTracking } from '@/hooks/useHoverTracking';
 import PriceDisplay from '@/components/ui/PriceDisplay';
 
@@ -21,7 +22,7 @@ const AmenityIcon = ({ name }: { name: string }) => {
         breakfast: 'Breakfast',
         spa: 'Spa',
     };
-    return <span className="feature-tag">{iconMap[name.toLowerCase()] || name}</span>;
+    return <span className="feature-tag">{iconMap[name.toLowerCase()] || name.replaceAll("_", " ")}</span>;
 };
 
 export default function HotelCard({ hotel }: { hotel: Hotel }) {
@@ -52,13 +53,24 @@ export default function HotelCard({ hotel }: { hotel: Hotel }) {
             className="hotel-card cursor-pointer"
             onClick={handleCardClick}
         >
-            <div className="hotel-image bg-gray-200 flex items-center justify-center">
+            <div className="hotel-image relative overflow-hidden">
+                <img
+                    src={getHotelImageUrl(hotel.id, { w: 400, h: 300 })}
+                    alt={hotel.name}
+                    className="w-full h-full object-cover"
+                    onError={(e) => {
+                        e.currentTarget.style.display = 'none';
+                        const fallback = e.currentTarget.nextElementSibling as HTMLElement;
+                        if (fallback) fallback.style.display = 'flex';
+                    }}
+                />
+                <div className="absolute inset-0 bg-gray-200 flex items-center justify-center" style={{ display: 'none' }}>
                     <span className="text-gray-400 text-sm">Image</span>
                 </div>
+            </div>
 
             <div className="hotel-info">
                 <h3 ref={titleRef} className="hotel-name">{hotel.name}</h3>
-                <div className="hotel-location text-sm mb-2">{hotel.roomType}</div>
                 <div className="text-sm text-[var(--text-secondary)] mb-2">
                     {hotel.checkIn} - {hotel.checkOut}
                 </div>
@@ -67,9 +79,6 @@ export default function HotelCard({ hotel }: { hotel: Hotel }) {
                         <AmenityIcon key={a} name={a} />
                     ))}
                 </div>
-                {hotel.refundable && (
-                    <div className="free-cancellation mt-2">Free cancellation</div>
-                )}
             </div>
 
             <div className="hotel-pricing">

web/src/components/feats/hotel/HotelDetails.tsx

@@ -1,6 +1,9 @@
 'use client';
 
+import { useState, useEffect } from 'react';
 import type { Hotel } from '@/lib/hotel-utils';
+import { getHotelImageUrl } from '@/lib/hotel-utils';
+import PriceDisplay from '@/components/ui/PriceDisplay';
 
 interface HotelDetailsProps {
   product: Hotel;
@@ -8,19 +11,61 @@ interface HotelDetailsProps {
   addedToCart: boolean;
 }
 
+const PriceTotalDisplay = ({ productId, nights }: { productId: string; nights: number }) => {
+  const [price, setPrice] = useState<number | null>(null);
+
+  useEffect(() => {
+    const fetchPrice = async () => {
+      try {
+        const sessionRes = await fetch('/api/session');
+        const sessionData = await sessionRes.json();
+        const params = new URLSearchParams({
+          productId,
+          sessionId: sessionData.sessionId || '',
+          experimentId: sessionData.experimentId || '',
+        });
+        const res = await fetch(`/api/pricing?${params.toString()}`);
+        const data = await res.json();
+        setPrice(data.price);
+      } catch (err) {
+        console.error('failed to fetch price for total:', err);
+      }
+    };
+    fetchPrice();
+  }, [productId]);
+
+  if (!price) return <span className="text-4xl font-bold text-gray-900">Loading...</span>;
+
+  return (
+    <span className="text-4xl font-bold text-gray-900">
+      ${(price * nights).toFixed(2)}
+    </span>
+  );
+};
+
 export default function HotelDetails({ product, onAddToCart, addedToCart }: HotelDetailsProps) {
   return (
     <div className="w-full flex flex-col lg:flex-row gap-12 py-8">
-      {/* Image Section - Larger and cleaner */}
+      <div className="w-full lg:w-1/2 rounded-lg aspect-[4/3] overflow-hidden shrink-0">
-      <div className="w-full lg:w-1/2 bg-gray-100 rounded-lg aspect-[4/3] flex items-center justify-center shrink-0">
+        <img
+          src={getHotelImageUrl(product.id, { w: 800, h: 600 })}
+          alt={product.name}
+          className="w-full h-full object-cover"
+          onError={(e) => {
+            e.currentTarget.style.display = 'none';
+            if (e.currentTarget.nextElementSibling) {
+              (e.currentTarget.nextElementSibling as HTMLElement).style.display = 'flex';
+            }
+          }}
+        />
+        <div className="w-full h-full bg-gray-100 rounded-lg flex items-center justify-center" style={{ display: 'none' }}>
           <span className="text-gray-400 text-lg font-medium">Hotel Image</span>
         </div>
+      </div>
 
-      {/* Details Section - Full height/width usage */}
       <div className="flex-1 flex flex-col">
         <div className="border-b pb-6 mb-6">
           <h1 className="text-4xl font-bold text-gray-900 mb-2">{product.name}</h1>
-          <p className="text-xl text-gray-500">{product.roomType}</p>
         </div>
 
         <div className="grid grid-cols-2 gap-8 mb-8">
@@ -39,24 +84,17 @@ export default function HotelDetails({ product, onAddToCart, addedToCart }: Hote
           <div className="flex flex-wrap gap-3">
             {product.amenities.map(a => (
               <span key={a} className="px-3 py-1.5 bg-gray-100 text-gray-700 rounded-md text-sm font-medium">
-                {a}
+                  {a.replaceAll('_', ' ')}
               </span>
             ))}
           </div>
         </div>
 
-        {product.refundable && (
-          <div className="mb-8 p-4 bg-green-50 text-green-800 rounded-md inline-block">
-            <span className="font-medium">Free cancellation available</span>
-          </div>
-        )}
-
         <div className="mt-auto pt-6 border-t flex items-center justify-between">
           <div>
-            <p className="text-sm text-gray-500 mb-1">Total for {product.nights} night{product.nights > 1 ? 's' : ''}</p>
+            <p className="text-sm text-gray-500 mb-1">Price per night</p>
-            <div className="flex items-baseline gap-2">
+            <div className="mb-3">
-              <span className="text-4xl font-bold text-gray-900">${product.pricePerNight * product.nights}</span>
+              <PriceDisplay productId={product.id} className="!text-2xl" />
-              <span className="text-gray-500">/ {product.nights} nights</span>
             </div>
           </div>
 

web/src/components/ui/DateInput.tsx

@@ -1,7 +1,29 @@
-import { InputHTMLAttributes } from 'react';
+import { InputHTMLAttributes, useMemo } from 'react';
 
 interface DateInpProps extends Omit<InputHTMLAttributes<HTMLInputElement>, 'type'> {}
 
 export default function DateInput({ className = '', ...props }: DateInpProps) {
-  return <input type="date" className={`input-field ${className}`.trim()} {...props} />;
+  const { minDate, maxDate } = useMemo(() => {
+    const today = new Date();
+    const tomorrow = new Date(today);
+    tomorrow.setDate(today.getDate() + 1);
+
+    const tenDaysOut = new Date(tomorrow);
+    tenDaysOut.setDate(tomorrow.getDate() + 9); // tomorrow + 9 = 10 days total
+
+    return {
+      minDate: tomorrow.toISOString().split('T')[0],
+      maxDate: tenDaysOut.toISOString().split('T')[0]
+    };
+  }, []);
+
+  return (
+    <input
+      type="date"
+      className={`input-field ${className}`.trim()}
+      min={minDate}
+      max={maxDate}
+      {...props}
+    />
+  );
 }

web/src/components/ui/Navigation.tsx

@@ -37,9 +37,7 @@ export default function Navigation() {
           <div className="flex items-center space-x-1">
             <NavLink href="/">Home</NavLink>
             <NavLink href="/products">Products</NavLink>
-            <NavLink href="/search">Search</NavLink>
             <NavLink href="/cart">Cart</NavLink>
-            <NavLink href="/checkout">Checkout</NavLink>
           </div>
         </div>
       </div>

web/src/components/ui/SelectDropdown.tsx

@@ -0,0 +1,119 @@
+'use client';
+
+import { useState, useRef, useEffect, ReactNode } from 'react';
+
+export interface SelectOption {
+  value: string;
+  label: string;
+  sublabel?: string;
+}
+
+interface SelectDropdownProps {
+  value: string;
+  onChange: (value: string) => void;
+  options: SelectOption[];
+  placeholder?: string;
+  icon?: ReactNode;
+  required?: boolean;
+  id?: string;
+}
+
+export default function SelectDropdown({
+  value,
+  onChange,
+  options,
+  placeholder = 'Select...',
+  icon,
+  required,
+  id,
+}: SelectDropdownProps) {
+  const [open, setOpen] = useState(false);
+  const [filter, setFilter] = useState('');
+  const ref = useRef<HTMLDivElement>(null);
+  const inputRef = useRef<HTMLInputElement>(null);
+
+  useEffect(() => {
+    const handleClick = (e: MouseEvent) => {
+      if (ref.current && !ref.current.contains(e.target as Node)) {
+        setOpen(false);
+        setFilter('');
+      }
+    };
+    document.addEventListener('mousedown', handleClick);
+    return () => document.removeEventListener('mousedown', handleClick);
+  }, []);
+
+  const selectedOption = options.find((o) => o.value === value);
+  const filtered = options.filter(
+    (o) =>
+      o.label.toLowerCase().includes(filter.toLowerCase()) ||
+      o.value.toLowerCase().includes(filter.toLowerCase()) ||
+      o.sublabel?.toLowerCase().includes(filter.toLowerCase())
+  );
+
+  const handleSelect = (opt: SelectOption) => {
+    onChange(opt.value);
+    setOpen(false);
+    setFilter('');
+  };
+
+  return (
+    <div className="relative" ref={ref}>
+      <div
+        className="input-field flex items-center gap-2 cursor-pointer box-border"
+        onClick={() => {
+          setOpen(true);
+          setTimeout(() => inputRef.current?.focus(), 0);
+        }}
+      >
+        {icon && <span className="text-[var(--text-secondary)]">{icon}</span>}
+        {open ? (
+          <input
+            ref={inputRef}
+            type="text"
+            id={id}
+            value={filter}
+            onChange={(e) => setFilter(e.target.value)}
+            placeholder={placeholder}
+            className="flex-1 bg-transparent outline-none text-sm text-[var(--text-primary)]"
+          />
+        ) : (
+          <span className={`flex-1 text-sm ${value ? 'text-[var(--text-primary)]' : 'text-[var(--text-secondary)]'}`}>
+            {selectedOption ? selectedOption.label : placeholder}
+          </span>
+        )}
+        <svg
+          className={`w-4 h-4 text-[var(--text-secondary)] transition-transform ${open ? 'rotate-180' : ''}`}
+          fill="none"
+          stroke="currentColor"
+          viewBox="0 0 24 24"
+        >
+          <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M19 9l-7 7-7-7" />
+        </svg>
+      </div>
+      {open && (
+        <div className="absolute z-20 mt-1 w-full bg-[var(--bg-primary)] border-2 border-[var(--accent-primary)] rounded-md shadow-lg max-h-60 overflow-y-auto">
+          {filtered.length === 0 ? (
+            <div className="px-4 py-3 text-sm text-[var(--text-secondary)]">No results</div>
+          ) : (
+            filtered.map((opt) => (
+              <div
+                key={opt.value}
+                onClick={() => handleSelect(opt)}
+                className={`px-4 py-2 cursor-pointer transition-colors hover:bg-[var(--accent-primary-light)] ${
+                  opt.value === value ? 'bg-[var(--accent-primary-light)]' : ''
+                }`}
+              >
+                <div className="text-sm font-medium text-[var(--text-primary)]">{opt.label}</div>
+                {opt.sublabel && <div className="text-xs text-[var(--text-secondary)]">{opt.sublabel}</div>}
+              </div>
+            ))
+          )}
+        </div>
+      )}
+      {required && !value && (
+        <input type="text" required className="sr-only" tabIndex={-1} value="" onChange={() => {}} />
+      )}
+    </div>
+  );
+}

web/src/components/ui/index.ts

@@ -5,3 +5,5 @@ export { default as DateInput } from './DateInput';
 export { default as RadioGroup } from './RadioGroup';
 export { default as Dropdown, DropdownCounter } from './Dropdown';
 export { default as Navigation } from './Navigation';
+export { default as SelectDropdown } from './SelectDropdown';
+export type { SelectOption } from './SelectDropdown';

web/src/lib/airline-utils.ts

@@ -31,7 +31,7 @@ export interface Flight {
   availability: number;
 }
 
-const EPOCH = new Date(0);
+import { dateToDaysFromToday, dateToIndex, todayIndex } from './date-utils';
 
 export const transformProduct = (p: AirlineProduct): Flight => {
   const { id, flight_type, date_index, metadata, availability } = p;
@@ -52,24 +52,4 @@ export const transformProduct = (p: AirlineProduct): Flight => {
   };
 };
 
-// convert date string to days from today
+export { dateToDaysFromToday, dateToIndex, todayIndex };
-export const dateToDaysFromToday = (dateStr: string): number => {
-  const target = new Date(dateStr);
-  target.setHours(0, 0, 0, 0);
-  const today = new Date();
-  today.setHours(0, 0, 0, 0);
-  return Math.floor((target.getTime() - today.getTime()) / 86400000);
-};
-
-// convert date string to date_index (days since epoch)
-export const dateToIndex = (dateStr: string): number => {
-  const d = new Date(dateStr);
-  return Math.floor((d.getTime() - EPOCH.getTime()) / 86400000);
-};
-
-// get current date_index
-export const todayIndex = (): number => {
-  const now = new Date();
-  now.setHours(0, 0, 0, 0);
-  return Math.floor((now.getTime() - EPOCH.getTime()) / 86400000);
-};

web/src/lib/config.ts

@@ -16,7 +16,7 @@ const envSchema = z.object({
 // parse and validate env at module load, fail fast with descriptive errors
 const parseEnv = (): Env => {
     const result = envSchema.safeParse({
-        STORE_MODE: process.env.STORE_MODE,
+        STORE_MODE: process.env.NEXT_PUBLIC_STORE_MODE || process.env.STORE_MODE,
         NEXT_PUBLIC_API_BASE: process.env.NEXT_PUBLIC_API_BASE,
         NEXT_PUBLIC_APP_ENV: process.env.NEXT_PUBLIC_APP_ENV,
     });

web/src/lib/date-utils.ts

@@ -0,0 +1,23 @@
+const EPOCH = new Date(0);
+const MS_PER_DAY = 86400000;
+
+export const dateToDaysFromToday = (dateStr: string): number => {
+  const target = new Date(dateStr);
+  target.setHours(0, 0, 0, 0);
+  const today = new Date();
+  today.setHours(0, 0, 0, 0);
+  return Math.floor((target.getTime() - today.getTime()) / MS_PER_DAY);
+};
+
+export const dateToIndex = (dateStr: string): number => {
+  const d = new Date(dateStr);
+  return Math.floor((d.getTime() - EPOCH.getTime()) / MS_PER_DAY);
+};
+
+export const todayIndex = (): number => {
+  const now = new Date();
+  now.setHours(0, 0, 0, 0);
+  return Math.floor((now.getTime() - EPOCH.getTime()) / MS_PER_DAY);
+};
+
+export { EPOCH, MS_PER_DAY };

web/src/lib/hotel-utils.ts

@@ -21,51 +21,78 @@ export interface Hotel {
   checkOut: string;
   dateIndex: number;
   amenities: string[];
-  refundable: boolean;
   pricePerNight: number;
   nights: number;
 }
 
-const EPOCH = new Date(0);
+import { EPOCH, MS_PER_DAY, dateToDaysFromToday, dateToIndex, todayIndex } from './date-utils';
 
 export const transformProduct = (p: HotelProduct): Hotel => {
   const { id, room_type, date_index, metadata } = p;
-  const checkIn = new Date(EPOCH.getTime() + date_index * 86400000);
+
+  // DB stores date_index as days since epoch
+  // but if value is small (<1000), treat as days from today for backward compat
+  let checkIn: Date;
+  if (date_index < 1000) {
+    // legacy: treat as offset from today
+    const today = new Date();
+    today.setHours(0, 0, 0, 0);
+    checkIn = new Date(today.getTime() + date_index * MS_PER_DAY);
+  } else {
+    // proper: days since epoch
+    checkIn = new Date(EPOCH.getTime() + date_index * MS_PER_DAY);
+  }
+
   const nights = 1;
-  const checkOut = new Date(checkIn.getTime() + nights * 86400000);
+  const checkOut = new Date(checkIn.getTime() + nights * MS_PER_DAY);
+
+  const formatOpts: Intl.DateTimeFormatOptions = {
+    month: 'short',
+    day: 'numeric',
+    year: checkIn.getFullYear() !== new Date().getFullYear() ? 'numeric' : undefined
+  };
 
   return {
     id,
     name: metadata?.name || room_type,
     roomType: room_type,
-    checkIn: checkIn.toLocaleDateString('en-US', { month: 'short', day: 'numeric' }),
+    checkIn: checkIn.toLocaleDateString('en-US', formatOpts),
-    checkOut: checkOut.toLocaleDateString('en-US', { month: 'short', day: 'numeric' }),
+    checkOut: checkOut.toLocaleDateString('en-US', formatOpts),
     dateIndex: date_index,
     amenities: metadata?.amenities || [],
-    refundable: metadata?.refundable || false,
     pricePerNight: metadata?.base_price || 100,
     nights,
   };
 };
 
-// convert date string to days from today
+const hotelImagePool = [
-export const dateToDaysFromToday = (dateStr: string): number => {
+  'photo-1566073771259-6a8506099945',
-  const target = new Date(dateStr);
+  'photo-1551882547-ff40c63fe5fa',
-  target.setHours(0, 0, 0, 0);
+  'photo-1590490360182-c33d57733427',
-  const today = new Date();
+  'photo-1582719478250-c89cae4dc85b',
-  today.setHours(0, 0, 0, 0);
+  'photo-1596701062351-8c2c14d1fdd0',
-  return Math.floor((target.getTime() - today.getTime()) / 86400000);
+  'photo-1631049307264-da0ec9d70304',
+  'photo-1578683010236-d716f9a3f461',
+  'photo-1540518614846-7eded433c457',
+  'photo-1505693416388-ac5ce068fe85',
+  'photo-1522771739844-6a9f6d5f14af',
+  'photo-1562438668-bcf0ca6578f0',
+  'photo-1595576508898-0ad5c879a061',
+];
+
+const hashString = (s: string): number => {
+  let h = 0;
+  for (let i = 0; i < s.length; i++) {
+    h = ((h << 5) - h) + s.charCodeAt(i);
+    h = h & h;
+  }
+  return Math.abs(h);
 };
 
-// convert date string to date_index (days since epoch)
+export const getHotelImageUrl = (hotelId: string, size: { w: number; h: number } = { w: 400, h: 300 }): string => {
-export const dateToIndex = (dateStr: string): number => {
+  const idx = hashString(hotelId) % hotelImagePool.length;
-  const d = new Date(dateStr);
+  const photoId = hotelImagePool[idx];
-  return Math.floor((d.getTime() - EPOCH.getTime()) / 86400000);
+  return `https://images.unsplash.com/${photoId}?w=${size.w}&h=${size.h}&fit=crop`;
 };
 
-// get current date_index
+export { dateToDaysFromToday, dateToIndex, todayIndex };
-export const todayIndex = (): number => {
-  const now = new Date();
-  now.setHours(0, 0, 0, 0);
-  return Math.floor((now.getTime() - EPOCH.getTime()) / 86400000);
-};

web/src/styles/airline.css

@@ -278,6 +278,8 @@
   padding: 12px;
   transition: border-color 0.2s ease;
   width: 100%;
+  min-height: 48px;
+  box-sizing: border-box;
 }
 
 [data-mode="airline"] .input-field:focus {