Keynote: From Platforms to AI Factories. Has Kubernetes Solved It? - Saiyam Pathak

This keynote explores how Kubernetes is evolving to support AI factory workloads, addressing the challenge of safely sharing expensive GPU infrastructure across multiple AI teams. It covers HAMi for virtualizing GPUs into smaller slices with individual memory budgets, and DRA for enabling Kubernetes to understand GPUs as rich devices rather than opaque numbers. The content is aimed at platform engineers and AI/ML infrastructure teams looking to optimize GPU utilization.