Aegis

A disaster response program useful for a post-disaster. Made for Hack the Coast 2026

Stack

• Frontend: React + TypeScript + Vite
• Backend: FastAPI + MongoDB
• AI Voice: ElevenLabs Conversational AI
• Hosting: Vultr (frontend & backend)

Local Setup

Frontend

cd frontend && npm install && npm run dev

Create frontend/.env:

VITE_API_URL=http://localhost:8000
VITE_ELEVENLABS_AGENT_ID=your-agent-id

Backend

See Quick Start below.

ElevenLabs Voice Call

Visit /call or use Voice Call from the "Need Help?" flow to speak with an AI agent. The agent collects your situation and location for responders.

Setup:

1. Create an agent at ElevenLabs Conversational AI
2. Copy the agent ID and add to frontend/.env:

   VITE_ELEVENLABS_AGENT_ID=agent_xxxxxxxx
   

The call log shows the full conversation transcript and persists after the call ends.

Vultr Production

Frontend and backend are hosted on Vultr. Configure the frontend to point at your API:

VITE_API_URL=http://YOUR_VULTR_IP/api

Replace YOUR_VULTR_IP with your server IP (e.g. 149.28.10.116).

🛡️ Aegis Backend

The central nervous system for the Aegis emergency response platform. This FastAPI-powered backend handles real-time incident ingestion, runs a weighted priority scoring algorithm, and serves as the data bridge between the frontend and MongoDB.

🚀 Quick Start

1. Prerequisites

• Python 3.9+
• MongoDB Atlas account (or a local MongoDB instance)

2. Installation

# Clone the repository
git clone [https://github.com/kevinli808/Aegis.git](https://github.com/kevinli808/Aegis.git)
cd Aegis

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

3. Environment Setup

Create a .env file in the root directory and add your MongoDB connection string:

MONGO_URI=mongodb+srv://<username>:<password>@cluster.mongodb.net/aegis

4. Run the server

uvicorn main:app --reload

🛠️ API Architecture

Priority Scoring Algorithm Every report sent to /report is processed through our scoring engine. It assigns a Priority Level (1-3) based on:

• Incident Type: Medical, Fire, and Trapped (High) vs. Power Outage (Low).
• Scale: Number of people affected.
• Status: Immediate danger or life-threatening symptoms (Unconscious, Breathing Difficulty).

Key Endpoints

Method	Endpoint	Description
POST	/report	Ingests new data and runs the scoring algorithm
GET	/incidents/stats	Returns active incidents. Use ?map_view=true for map markers.
GET	/incidents	Returns active incidents by priority level.
PATCH	/incidents/{incident_id}/resolve	Updates an incident status from "active" to "resolved"
GET	/incidents/{incident_id}	Returns incident details by incident ID.

📁 Project Structure

• main.py: Application entry point, CORS settings, and route definitions.
• database.py: MongoDB connection and client setup.
• crud.py: Database helper functions for creating and updating incidents.
• seed_data.py: Script to populate the database with mock Vancouver-based incidents.
• wipe.py: Script to clean all incident data in the MongoDB.