In this advanced 400-level workshop, you will learn how to optimize Amazon EKS and Karpenter for cost performance when running Apache Spark jobs and visualize job-level costs to inform decisions around migration, expansion, and ongoing optimization. Throughout the workshop, you will run the same Spark job five times. Before each run, you will apply different infrastructure configurations, including Spot instances, NVMe storage, and Graviton, and compare how quickly the job completes. In the final section, you will monitor and analyze cost per job. By the end of the workshop, you will be able to optimize analytics workloads on Spark on Amazon EKS, compare infrastructure options, and make informed cost-performance decisions.