Air-gapped MLOps on Kubernetes: MLflow, DVC, and OpenBao

A DevOps-oriented blueprint for training and serving models behind the perimeter—GitOps, secrets, and GPU outside the cluster.