Run:ai

NVIDIA Run:ai is an open-source AI workload and GPU orchestration platform designed for Kubernetes environments. It provides centralized management, resource scheduling, and observability to help organizations efficiently share GPU resources across development, training, and inference workloads.

NVIDIA Run:ai is an open-source AI workload and GPU orchestration platform designed for Kubernetes environments. It provides centralized management, resource scheduling, and observability to help organizations efficiently share GPU resources across development, training, and inference workloads.

• 01AI-aware GPU scheduling with fair-share, preemption, and priority support
• 02Dynamic fractional GPU allocation to maximize resource utilization
• 03Multi-cluster management for centralized control across hybrid infrastructure
• 04Comprehensive observability for monitoring nodes, GPUs, and workload performance
• 05Role-based access control and organizational policy enforcement
• 06Model Context Protocol (MCP) server for integration with AI assistants