This article announces Spark Connect support on Amazon EMR Serverless, enabling interactive PySpark development from local environments while executing at scale on EMR Serverless.

• Develop and debug Spark applications locally in IDEs, Jupyter notebooks, or SageMaker without deploy-and-check cycles
• Client-server architecture separates application code from Spark engine via secure gRPC/TLS connection
• Each session has unique ARN enabling per-session IAM permissions, cost allocation, and CloudTrail auditing
• Pay only for compute resources during active sessions; automatic scaling via dynamic resource allocation
• Supports interactive ETL, dbt-spark, Iceberg analytics, S3 Tables, and exploratory data analysis
• Available in EMR release 7.13.0 (Spark 3.5.6) and later across all AWS regions
• No additional charge; pricing matches standard EMR Serverless batch job model

Spark Connect on EMR Serverless simplifies PySpark development by bridging local development and production-scale execution with automatic infrastructure management and fine-grained access controls.