Analytics
AWS Glue
AWS Glue is a serverless data integration service for discovering, cataloging, transforming, and validating data. GenAI teams use it to prepare governed corpora while Glue Data Quality evaluates rules and publishes quality outcomes.
Key points
- Crawlers and the Data Catalog make schemas discoverable
- Spark and Python shell jobs support managed transformations
- Glue Data Quality uses rulesets to identify and quarantine unsuitable records
When to use it
- Clean and partition documents before RAG ingestion
- Reject malformed or incomplete training and evaluation records
Exam tips
- Select Glue for repeatable data-lake ETL; Lambda is better for small event-level transforms
- A passed schema check does not prove semantic fitness, so retain lineage and domain quality tests