AeroGuard is a cinematic, AI-driven atmospheric intelligence platform designed to transform invisible air pollution data into actionable health insights. Built for the AIColegion Hackathon 2026 (VESIT), it combines real-time sensory data with machine learning to predict, explain, and guard against environmental threats across India.
- 6-Hour AQI Forecast: Uses machine learning to predict atmospheric trajectory.
- Explainable AI (X-AI): Transparently shows which pollutants (PM2.5, NO₂, etc.) are driving the forecast using feature importance scores.
- Persona-Aware Assessments: Tailored health logic for the General Public, Children/Elderly, and Athletes.
- AI Briefings: Human-readable explanations of current conditions and protective protocols.
- Nationwide Coverage: Real-time AQI visualization across 45+ major Indian cities and a high-density 10x10 regional grid scan.
- Smooth Gradient Rendering: Advanced Heatmap implementation for a professional, "vibrant" aesthetic.
- Geographically Bounded: Precisely calibrated for the Indian subcontinent.
- Historical Deep-Dives: 7-to-30 day trend analysis for multi-pollutant tracking.
- Intelligent Search: Fuzzy location matching with geometric bounding for India-only results.
- API Constraints: Currently relies on the WAQI API which has a default programmatic quota of 1,000 requests per minute (with bursts up to 60 requests). Full-scale commercial deployment would require a dedicated institutional agreement.
- Hosting: Designed as a local-first development project; scalable cloud hosting (AWS/GCP) for the ML models and data ingestion is planned for post-competition.
- Sensor Density: While we use 250+ official stations, hyper-local street-level data is currently limited by public station availability.
- 📍 DIY IoT Integration: Support for low-cost PM2.5 sensors (ESP32/Arduinos) for community-driven data.
- 📱 Mobile Ecosystem: Cross-platform Flutter/React Native app with "High Pollution" push alerts based on live location.
- 🌦️ Met-AI Sync: Integrating real-time wind speed and humidity from OpenWeatherMap to improve 24h forecasting.
AeroGuard/
├── Backend/ # Flask / FastAPI Architecture
│ ├── app/
│ │ ├── routes/ # API Blueprints (AI, AQI, Forecast, Risk)
│ │ ├── services/ # Business Logic & ML Service Layer
│ │ └── utils/ # Shared helpers & Error Handlers
│ └── run.py # Main Entry Point
├── frontend/ # React / Vite Infrastructure
│ ├── src/
│ │ ├── Components/ # Complex UI (Heatmap, Analytics, Search)
│ │ ├── pages/ # High-level Views (Dashboard, Risk, Landing)
│ │ └── api/ # Frontend API Utilities
│ └── package.json
└── Readme.md
GET /api/v1/realtime-aqi/city/<city_name>: Fetch live data for a city.GET /api/v1/realtime-aqi/nationwide: High-density data points for the heatmap.GET /api/v1/realtime-aqi/token: Securely proxy WAQI tokens to the frontend.
GET /api/v1/ai/briefing?city=...&persona=...: Generate personalized AI health advice.POST /api/v1/ai/explain-forecast: ML feature importance explanation.
GET /api/v1/health-risk: Multi-factor persona-based risk assessment.
cd Backend
python -m venv venv
# Windows
.\venv\Scripts\activate
# Dependencies
pip install -r requirements-fixed.txt
# Run
python run.pyNote: Ensure REALTIME_WAQI_API_KEY is set in your .env.
cd frontend
npm install
npm run devBuilt with ❤️ by Team 70 — CultBoyz for AIColegion 2026 @ VESIT (Vivekanand Education Society's Institute of Technology).
 Anshul Patil Frontend Design & API Integration |
 Archit Chitte Backend, ML Models & Model Integration |
© 2026 AeroGuard Intelligence. Part of the CultBoyz hackathon suite.