Explain How Kubernetes Works With GPU Like I’m 5 - Carlos Santana, AWS

Carlos Santana explains how Kubernetes orchestrates GPU workloads, breaking down the integration stack using NVIDIA Jetson hardware.

Don't miss out! Join us at our next Flagship Conference: KubeCon + CloudNativeCon events in Hong Kong, China (June 10-11); Tokyo, Japan (June 16-17); Hyderabad, India (August 6-7); Atlanta, US (November 10-13). Connect with our current graduated, incubating, and sandbox projects as the community gathers to further the education and advancement of cloud native computing. Learn more at kubecon.io Explain How Kubernetes Works With GPU Like I’m 5 - Carlos Santana, AWS Want to understand how Kubernetes handles GPUs? Join us for a beginner-friendly deep dive into GPU integration using a homelab setup with NVIDIA Jetson hardware. Rather than relying solely on operators, we'll break down the entire stack to reveal how Kubernetes orchestrates GPU workloads. This session walks through the complete journey of enabling GPU support in Kubernetes, from bare metal to running GPU-accelerated containers. Using a practical homelab example with a Jetson NUC, we'll explore how Kubernetes detects and manages GPU hardware, the critical role of drivers and container toolkit, and how kubelet plugins enable GPU support. You'll understand the mechanics of node labeling, GPU resource allocation, and the process of requesting GPU resources in Pod specifications. We'll also demystify CUDA and its essential role in GPU computing. Whether you're new to GPU computing or looking to understand the internals beyond operator abstractions, this talk is for you.