From local to live in 10 minutes

Modern Python deployment isn't about renting a VPS and configuring nginx. Three platforms handle everything: Fly.io, Railway, Render. Vercel is best for Next.js + Python serverless functions.

The three files you need

1. Dockerfile — how to build your app. Same file works everywhere.
2. requirements.txt — pinned deps (pip freeze > requirements.txt).
3. Platform config — fly.toml, railway.json, or render.yaml.

Deploying to Fly.io (recommended for India — Mumbai region bom)

brew install flyctl        # or curl -L https://fly.io/install.sh | sh
fly auth signup
fly launch                 # interactive wizard, generates fly.toml
fly deploy                 # ships it

Get a URL like https://my-python-api.fly.dev. Free tier: 3 shared-cpu-1x machines.

Deploying to Railway

npm i -g @railway/cli
railway login
railway init
railway up

Even simpler — auto-detects Python, no Dockerfile needed for most apps.

Environment variables

Never commit secrets. Set them per-platform:

• Fly: fly secrets set DATABASE_URL=postgres://...
• Railway: dashboard or railway variables set
• Locally: .env file + python-dotenv

Add a database

• Fly Postgres: fly postgres create — free 1GB.
• Neon: serverless Postgres, generous free tier, works with any host.
• Supabase: Postgres + auth + storage + realtime — batteries included.

The 3 things that always break

1. Wrong port — platform expects 8080, you hardcoded 5000. Bind to 0.0.0.0, not 127.0.0.1.
2. Missing env var — app crashes on boot with KeyError.
3. pip deps compile from source — pin python:3.13-slim and use pip install --no-cache-dir.

Try it

1. Write a minimal FastAPI app locally.
2. Copy the Dockerfile above, run fly launch, deploy.
3. Hit your .fly.dev URL in the browser.