A full-stack social feed prototype where users control the recommendation algorithm.
- React + Vite frontend
- Python FastAPI backend
- SQLite database
cd server
python -m venv .venv
source .venv/bin/activate # Windows Git Bash: source .venv/Scripts/activate
pip install -r requirements.txt
uvicorn main:app --reload --host 0.0.0.0 --port 8000cd client
npm install
cp .env.example .env
npm run devOpen http://localhost:5173.
Create a new Web Service from this repo.
- Root Directory:
server - Build Command:
pip install -r requirements.txt - Start Command:
uvicorn main:app --host 0.0.0.0 --port $PORT - Environment variable:
CORS_ORIGINS=https://your-vercel-domain.vercel.app
For MVP SQLite works, but Render free instances may reset disk state. For durable production data, swap SQLite for Postgres.
- Root Directory:
client - Build Command:
npm run build - Output Directory:
dist - Environment variable:
VITE_API_URL=https://your-render-service.onrender.com
Move the algorithm sliders and the frontend calls /api/feed/rank. The backend reorders posts based on metadata scores and returns explanations for why each post appears.