Explore the design and implementation of a high-performance Kubernetes cluster optimized for machine learning and AI workloads in an academic setting. Learn about the integration of NVIDIA DGX A100 machines for exceptional compute density and performance, as well as the use of open-source software for node integration. Discover the challenges and successes in incorporating external components such as CEPH storage, GitLab registry and runners, and SAML authentication. Gain insights into how the University of Alabama at Birmingham leverages container-enabled GPUs for research and development, addressing needs from regular ML training runs to supporting software development through CI pipelines. Understand how Kubernetes helps meet the growing demand for ad hoc, GPU-enabled compute capacity with complex software environments to power cloud-native workflows in academic research.
Kubernetes for GPU-Powered Machine Learning Workloads in Academia - Lecture
CNCF [Cloud Native Computing Foundation] via YouTube
Overview
Syllabus
Kubernetes For GPU Powered Machine Learning Workloads In... - Camille Rodriguez & John-Paul Robinson
Taught by
CNCF [Cloud Native Computing Foundation]