Explore a comprehensive conference talk on monitoring GPU clusters for AI/ML and HPC workloads at scale. Learn how NVIDIA addresses the monitoring needs of various user personas, including AI/ML researchers, operations teams, and stakeholders. Discover the combination of open-source tools used to meet diverse requirements and gain insights into deployment, maintenance, security, and scalability challenges encountered when monitoring GPU data. Understand how NVIDIA overcame these obstacles to create an effective monitoring solution for large GPU Kubernetes clusters running deep learning training workloads.