Operations

Runbooks for installing, reproducing, and deploying morel. The audience is operators and engineers shipping morel in production.

Detailed install options, dependency manifest, and the pinning policy for reproducibility.

Re-run a saved experiment deterministically from its config and manifest.

Docker, docker-compose, and a Kubernetes skeleton for the inference server.

Logging, Prometheus scraping, and the metrics the inference server exports.

Prometheus alert starters, triage guidance, and a repeatable model-bundle rollback procedure.

One-time Pages setup, deployment workflow, and the URL hygiene story.