Leverage Kubernetes To Optimize the Utilization of Your AI Accelerators

This session provides techniques to leverage Kubernetes to analyze the current utilization of AI accelerators and provides numerous tactics to optimize AI accelerator utilization for training, inference, notebooks, and other AI workloads. Nathan Beach, Product Manager at Google, discusses how the cost of AI accelerators like GPUs can be significant in AI model training and serving, and how Kubernetes can be used to manage and optimize their usage.