AidConnect 🚨

AI-Powered Disaster Response & Resource Coordination Platform

AidConnect is an AI-powered disaster response platform that intelligently matches shelter resources with affected areas during emergencies. Built with NVIDIA AI technology.

✨ Key Features

• 🤖 AI-Powered Resource Matching - NVIDIA Llama 3.3 Nemotron intelligently matches shelters with affected areas
• 🌦️ Weather Forecasting - NVIDIA FourCastNet for advanced weather simulation
• 🗺️ Interactive Maps - Real-time visualization of shelters, resources, and affected areas
• 💬 AI Assistant - Chat interface for disaster coordination
• 📊 Live Dashboard - Real-time metrics and alert feed

🏗️ Architecture

Frontend (React + TypeScript)
    ↓
Backend APIs (Python Flask)
    ↓
NVIDIA AI Models (Llama 3.3 + FourCastNet)

Components:

• Frontend: React + TypeScript + Tailwind CSS
• Resource Matching API: Flask + NVIDIA Llama 3.3
• Weather Simulation: NVIDIA FourCastNet
• MCP Hub: Multi-service orchestration

🔧 Prerequisites

• Node.js 18+ and npm
• Python 3.9+
• NVIDIA NGC API Key (Get it here)

📦 Installation

# Clone repository
git clone https://github.com/JeevaSaravanan/AidConnect.git
cd AidConnect

# Install frontend
npm install

# Install backend
python -m venv venv
source venv/bin/activate  # Windows: .\venv\Scripts\activate
cd mcp-hub && pip install -r requirements.txt
cd ../server/simulation && pip install -r requirements.txt

⚙️ Configuration

Create .env in mcp-hub/:

NV_API_KEY=your_nvidia_api_key
NGC_API_KEY=your_nvidia_ngc_api_key
NV_INVOKE_URL=https://integrate.api.nvidia.com/v1/chat/completions
NV_MODEL=nvidia/llama-3.3-nemotron-super-49b-v1.5

🚀 Running the Application

Terminal 1 - Frontend:

npm run dev

Terminal 2 - Resource Matching API:

cd mcp-hub
source venv/bin/activate
python flask_api.py

Terminal 3 - Weather Simulation:

cd server/simulation
source ../../venv/bin/activate
python app.py

Access the app at http://localhost:8080

📚 API Documentation

Resource Matching API (Port 5002)

POST /api/match-resources - Match affected areas with shelters using AI

GET /api/health - Health check

Weather Simulation API (Port 5001)

POST /forecast - Generate weather forecast videos

GET /video/<timestamp>/<variable> - Retrieve forecast video

For detailed API docs, see mcp-hub/MATCH_RESOURCES_README.md

📁 Project Structure

AidConnect/
├── src/                    # Frontend (React + TypeScript)
│   ├── components/        # UI components
│   ├── pages/            # Page routes
│   └── hooks/            # Custom hooks
├── mcp-hub/              # Backend APIs
│   ├── flask_api.py     # Resource matching API
│   ├── match_resources_api.py
│   └── llm_utils.py     # NVIDIA AI integration
├── server/simulation/    # Weather forecasting
│   └── app.py           # FourCastNet service
└── data/                # Shelter & volunteer data

🛠️ Technology Stack

Frontend: React 18, TypeScript, Vite, Tailwind CSS, shadcn/ui, Leaflet Maps

Backend: Python, Flask, FastAPI, FastMCP

AI/ML: NVIDIA Llama 3.3 Nemotron, NVIDIA FourCastNet, LangChain, RAG

Data: ArcGIS, FEMA API, OpenStreetMap

💻 Development

# Frontend development
npm run dev

# Linting
npm run lint

# Build for production
npm run build

🙏 Acknowledgments

• NVIDIA for AI APIs and FourCastNet technology
• shadcn/ui for component library
• OpenStreetMap, FEMA, and ArcGIS for data

📞 Contact

Jeeva Saravanan

Repository: github.com/JeevaSaravanan/AidConnect

---

🏆 Winner of NVIDIA Hackathon | Built with ❤️ for disaster response