Two production apps run on

Two production apps run on a five-year-old laptop in my room. Nginx handles TLS and routes subdomains to Docker containers; PostgreSQL sits behind them; an Ollama model runs locally as a fallback for the AI features.

I built it this way because managed platforms hide the layer where things actually break. Doing my own deploys, certs, and recovery means that when a client's app goes down in production, I've already debugged that layer — on hardware I couldn't just re-provision.