Install Ollama as a systemd service, configure production-ready settings, pull your first model, and verify the API is serving.