Kubernetes-based platform for GPU allocation, scheduling, partitioning and monitoring across AI training and inference workloads.