Data architectures face intense pressure as AI agents require rapid, concurrent queries over fresh data. In this technical session, discover how to run high-performance Spark jobs on Apache Iceberg data using Google Cloud. Learn how to deploy zero-ops serverless batch ETL pipelines, run ephemeral interactive sessions for PySpark data prep in your IDE of choice, and use agentic AI capabilities to build pipelines and troubleshoot errors. Explore how Lightning Engine bypasses JVM bottlenecks with native C++ execution and the Lakehouse runtime catalog enables zero-copy BigQuery analytics without lock-in.