A Spark job that takes records straight from the failed enriched good directory and deletes exactly those from DynamoDB - View it on GitHub