Explain how AWS Glue works and how to write a Python script for a serverless ETL job.

AWS Glue is a fully managed, serverless ETL service that automates data discovery, schema versioning, and job execution. A typical Glue workflow starts with a crawler that scans data sources (S3, JDBC, etc.) and populates the Glue Data Catalog. You then create a Glue job, which can be written in Python or Scala and runs on a managed Spark cluster. The job script reads data from the catalog, applies transformations using Spark DataFrames, and writes the output back to S3 or another target. Glue handles provisioning, scaling, and monitoring of the underlying cluster, so you pay only for the compute time used. You can schedule jobs with triggers, use job bookmarks to process incremental data, and integrate with AWS Step Functions for complex workflows. Writing a Python script for Glue involves importing the GlueContext, creating a DynamicFrame, performing transformations, and committing the result. This serverless model simplifies ETL development and maintenance.