Run

Run is a platform for optimizing and managing AI infrastructure, with a focus on getting more value from GPU resources. Built on Kubernetes, it helps teams allocate compute efficiently across users, projects, and workloads, and it requires basic Kubernetes knowledge to deploy and operate.

Key capabilities

• Dynamically distribute AI workloads across users and projects
• Monitor infrastructure utilization, GPU usage, and user activity
• Create configurable workspaces with selected tools and frameworks
• Manage quotas and access policies with flexible controls
• Schedule jobs with an AI workload scheduler
• Use GPU fractioning to run multiple tasks on a single GPU

Where it fits

Run integrates with both cloud and on-prem environments, supports multi-user operation, and scales for large AI and machine learning infrastructure. A running Kubernetes cluster is required.

Typical users

• Research labs and compute centers
• Companies running large AI projects
• Teams operating shared ML infrastructure