Back to catalog / Agent workflows

Create a Glue ETL job from S3 JSON to Parquet

Use the AWS data-processing MCP to discover S3 data, generate a PySpark transform, create Glue catalog/job resources, and verify with Athena.

Goal

Build reusable ETL from S3 JSON to Parquet for test data and verify that Athena can query the result.

In an isolated AWS test account, specify the input S3 prefix, output bucket, and Glue database. [src10-strands-create-s3-to-parquet-glue-etl-pipeline]
Have the agent inspect the schema and generate a Glue 5/PySpark script; review partitions, format, and role before creating the Glue job. [src10-strands-create-s3-to-parquet-glue-etl-pipeline]
Run the test job, confirm Parquet objects and the Glue table exist, then run a bounded Athena SELECT and inspect cost. [src10-strands-create-s3-to-parquet-glue-etl-pipeline]

Verify Glue job succeeded, output schema matches the target table, and Athena returns expected rows with bounded scanned data.

For job failure, inspect CloudWatch errors and IAM/S3 paths; if Athena cannot see the table, inspect Glue database and catalog/crawler updates.

A test AWS account, S3 test data, Bedrock model access, and least-privilege Glue/Athena/S3 roles.

Creates or changes Glue databases/tables/jobs, writes scripts and outputs to S3, and may incur Glue/Athena/EMR charges; use a test account and data.

Tools can create cloud resources; estimate costs, plan cleanup, and verify billing after removing sample resources.

Repository: strands-agents/samples
File path: python/04-industry-use-cases/software-engineering/amazon-dataprocessing-agent/README.md
Pinned commit: bd46f959844424a9544133e1263c8eae61e7fe9b
Source checked: 2026-10-14 · Reviewed against the pinned upstream source

These values describe the upstream repository, not popularity or likes for an individual skill or playbook.

No metrics snapshot is available yet.