Primarily a Python/PostGIS GIS dev, I’m moving more work to cloud-native data (COG, Parquet) and want a structured track that forces real code: PySpark/Databricks for ETL vs deepening SQL/PostGIS performance, or going TypeScript with MapLibre/deck.gl for front-end mapping. If you’ve taken a course or micro-credential in the last year that improved your commits (not just slides), which programs would you recommend for about 5 hours/week over 10–12 weeks?
Took the Databricks Data Engineer Associate labs last quarter and built a PySpark ETL that lands GeoParquet with Apache Sedona; the Auto Loader + Delta tests “forced real code” and replaced a pile of one-off GDAL glue. If you mostly ship data plumbing, this pays off quicker than going deep on MapLibre/deck.gl, but reserve a sprint to learn deck.gl’s layer patterns so you can debug the last mile with your front-end. https://academy.databricks.com/.
Building on @mason33, the combo that really forced code for me was Databricks Auto Loader + Delta Live Tables with Apache Sedona and Mosaic (H3): DLT expectations caught invalid geometries and we wrote partitioned GeoParquet keyed by H3 for fast S3 reads into MapLibre. Minor caveat: DLT adds cost, so prototype the pipeline as a standard job first.