This session focuses on performance and cost optimization on Amazon SageMaker. First, we'll show you how to automatically tune hyper-parameters, and quickly converge to optimal models. Second, you'll learn how to use SageMaker Neo, a new service that optimizes models for the underlying hardware architecture. Third, we'll show you how Elastic Inference lets you attach GPU acceleration to EC2 and SageMaker instances at the fraction of the cost of a full-fledged GPU instance. Finally, we'll share additional cost optimization tips for SageMaker.