An open-source distributed operating system built on top of Kubernetes that extends it across the full infrastructure stack, automatically handling the production configuration of bare-metal GPU clusters for AI inference and training.