Real-Time Multi-Agent Disaster Response Engine

A real-time disaster simulation platform that uses multiple AI agents to analyze risk zones, allocate emergency resources, and stream live decisions to a monitoring dashboard.

📌 Problem

Emergency response systems often struggle with:

• delayed situational awareness
• conflicting resource decisions
• lack of live coordination across teams

📺 Live Simulation Feed

Multi-Agent Flow & Spatial Rendering

🏗️ My Contribution

• designed multi-agent orchestration with LangGraph
• built real-time WebSocket event streaming
• implemented conflict resolution between resource-planning agents
• created route safety filtering over blocked road segments
• persisted simulation snapshots in MongoDB
• built live geospatial dashboard using React + Leaflet

🚀 How to Run locally

1. Database & Environment

You will need an active MongoDB connection and an OpenAI API Key.

# Set in your environment or .env
OPENAI_API_KEY=sk-...
MONGO_URI=mongodb://localhost:27017

(Note: A fallback mock mode executes locally if API keys drop).

2. Start the Backend (Engine)

cd backend
python -m venv venv
.\venv\Scripts\Activate.ps1
pip install -r requirements.txt
uvicorn app.main:app --host 0.0.0.0 --port 8000

3. Start the Frontend (Command Center Dash)

cd frontend
npm install
npm run dev

📊 Measurable Outputs

• Latency: average decision latency in mock mode is ~350ms per multi-agent cycle.
• Simulation: Continually processes multi-node road networks and shelter availability snapshots.
• Data Pipeline: Supports concurrent WebSocket clients receiving broadcasted MongoDB JSON states.

⚠️ Limitations

• uses simulated disaster feeds
• route planning currently works on a simplified graph
• no live government data feed integration yet
• LLM reasoning can vary, so deterministic fallbacks are included

---

Author: Pavan (Sai Pavan)