Hyperlocal Air Quality Forecasting & Attribution Platform
AQI Sentinel is an AI-powered Urban Air Quality Intelligence platform designed for city administrators. Moving beyond simple reactive dashboards, this platform provides predictive forecasting, geospatial source attribution, and actionable enforcement intelligence to help cities reduce pollution at its source.
🌐 Live Demo: https://etaihackathon.vercel.app/
Side-by-side comparison of Delhi (left) and Mumbai (right) showcasing dynamic intelligence adapting to specific city characteristics.
- Hyperlocal Forecasting (72H): Predicts AQI up to 3 days in advance using LightGBM models trained on historical and meteorological data.
- Dynamic Source Attribution: Uses SHAP values cross-referenced with ground-truth emission inventories (like UrbanEmissions.info APnA) to separate weather impacts from actual human-made pollution sources (e.g., Vehicle Exhaust vs. Industrial).
- Intervention Recommendation Engine: Automatically generates actionable insights and advisories (e.g., "Implement heavy-vehicle restrictions") based on the dominant pollution source in a specific grid.
- Multi-City Scalability: Architecture is designed to easily onboard new cities. Currently prototyped for Delhi, Mumbai, and Bengaluru.
- Frontend: React, Vite, Leaflet (Maps), Recharts
- Backend: Python, FastAPI, Uvicorn
- AI/ML: LightGBM, SHAP, Pandas, Scikit-Learn
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python3 -m uvicorn api.main:app --reload --env-file .env(Backend runs on http://localhost:8000)
cd frontend
npm install
npm run dev(Frontend runs on http://localhost:5173)
This project directly addresses the need for actionable intelligence. By showing exactly what is causing the pollution and recommending what to do about it, AQI Sentinel empowers policymakers to make data-driven decisions that actively improve public health.