A Kubernetes operator that provides smarter GPU scheduling — with time-slicing support, priority-based preemption, and utilization-aware placement.