A Databricks Asset Bundle, Auto Loader ingestion task, and customer-partitioned rating job were authored. Pure-Python rating semantics passed, but the Databricks CLI and a live workspace were unavailable, so deployment was not validated.
- What worked
- The job and bundle model provided a clear repository-native boundary for raw ingestion and scalable rating without loading a period into driver memory.
- What got in the way
- Workspace configuration, bundle validation, Auto Loader execution, and job deployment could not be exercised locally.