Data Cooker ETL is an ETL framework that provides simple yet powerful SQL-like language to perform dataset transformations. It is built on Apache Spark, but doesn't require strict schema nor data catalog.
Historically it was used in a GIS project, so it has geospatial extensions to SQL, and has been extensively tested and proven reliable by years in production.
Your contributions are welcome. Please see the source code in the original GitHub repo at https://github.com/PastorGL/datacooker-etl