Full-stack application that takes a street address or lat/long and returns a comprehensive property profile using Google Maps imagery, AI vision analysis, and geospatial data.

• Frontend: React + Vite, shadcn/ui, Tailwind CSS, react-map-gl, Recharts, TanStack Query
• Backend: Python FastAPI, OpenAI Vision (gpt-4o), SAM2 segmentation, Supabase
• Infrastructure: Docker Compose

• Node.js 20+
• Python 3.13+
• uv (Python package manager)
• Docker (optional)

git clone <repo-url>
cd Argus

cp .env.example .env
# Fill in your API keys in .env

make dev

This installs all dependencies and starts both the backend (http://localhost:8000) and frontend (http://localhost:5173) in parallel.

Other commands:

docker compose up --build

Full property analysis pipeline. Fetches imagery, runs AI analysis, computes isochrones, and finds nearby POIs.

Request:

{
  "address": "123 Main St, Toronto, ON",
  "lat": 43.6532,
  "lng": -79.3832
}

Response: Complete property profile including images, AI analysis, and nearby points of interest.

Fetch Street View and Satellite images for a location.

Query parameters: lat, lng, address

Response: URLs or base64-encoded images for the requested location.

Run AI vision analysis on property images.

Request:

{
  "images": ["<base64 or URL>"],
  "lat": 43.6532,
  "lng": -79.3832
}

Response: Structured AI analysis of the property (building type, condition, features, etc.).

Find nearby points of interest via OpenStreetMap/Overpass.

Query parameters: lat, lng, radius (metres), category

Response: List of nearby POIs with name, category, and distance.

Health check endpoint.

Response:

{ "status": "ok" }

Argus/
├── backend/
│   ├── app/
│   │   ├── __init__.py
│   │   ├── config.py
│   │   ├── models/
│   │   │   ├── __init__.py
│   │   │   └── schemas.py
│   │   └── services/
│   │       └── __init__.py
│   ├── Dockerfile
│   └── pyproject.toml
├── frontend/               # React + Vite app (to be scaffolded)
├── .env.example
├── .gitignore
├── docker-compose.yml
└── README.md