Short description (≤350 chars): PySpark analytics pipeline: ingest orders/products/customers/returns, transform with broadcast joins and windowing, data quality ...