How hirdeshpal888527/vantage works
Vantage is an A/B experimentation platform with a dual-write feature store. Experiment owners design tests through the dashboard or API; the API sizes experiments, assigns users, records metric events, and computes monitored results. A separate feature workflow registers and writes features, then serves online or point-in-time offline lookups. The README documents PostgreSQL and Redis roles; sampled code confirms the statistical modules and their API use. Some API internals were omitted from excerpts, so the graph avoids inferring their unshown wiring.
Main entry points: main.py FastAPI Service · feature_store.py Feature Store · app.py Streamlit Dashboard · assignment.py Variant Assignment
Components 16
User Interfaces
- Streamlit Dashboard
dashboard/app.py - FastAPI Service
api/main.py
Experiment Runtime
- Experiment Design
api/main.py - Variant Assignment
api/assignment.py - Metric Tracking
api/main.py - Experiment Results
api/main.py
Statistical Methods
- Power Analysis
stats/power_analysis.py - SRM Check
api/assignment.py - Sequential Test
stats/sequential_testing.py - CUPED Adjustment
stats/cuped.py - Bootstrap Intervals
stats/bootstrap.py
Feature Store
- Feature Endpoints
api/main.py - Feature Store
api/feature_store.py
Data Stores
- PostgreSQL
- Redis
Other components
- Experiment Owner
Connections 22
- Experiment Owner → Streamlit Dashboard: uses
README.md - Experiment Owner → FastAPI Service: calls
README.md - Streamlit Dashboard → FastAPI Service: requests
dashboard/app.py - FastAPI Service → Experiment Design: handles
api/main.py - Experiment Design → Power Analysis: computes size
api/main.py - Experiment Design → PostgreSQL: writes experiment
api/main.py - FastAPI Service → Variant Assignment: assigns users
api/main.py - Variant Assignment → PostgreSQL: stores assignment
- FastAPI Service → Metric Tracking: handles events
api/main.py - Metric Tracking → PostgreSQL: writes events
api/main.py - FastAPI Service → Experiment Results: computes results
api/main.py - Experiment Results → PostgreSQL: reads experiment data
api/main.py - Experiment Results → SRM Check: checks split
api/main.py - Experiment Results → Sequential Test: tests outcomes
api/main.py - Experiment Results → CUPED Adjustment: adjusts metrics
api/main.py - Experiment Results → Bootstrap Intervals: estimates intervals
api/main.py - FastAPI Service → Feature Endpoints: serves features
api/main.py - Feature Endpoints → Feature Store: delegates
api/main.py - Feature Store → PostgreSQL: writes history
api/feature_store.py - Feature Store → Redis: writes cache
api/feature_store.py - Feature Store → Redis: reads online
api/feature_store.py - Feature Store → PostgreSQL: queries as-of
api/feature_store.py