Design an idempotent churn ETL pipeline

This exercise assesses a candidate’s ability to design reliable ETL pipelines: idempotent processing, late-arriving streaming subscription events, SCD Type 2 dimensions, incremental backfills, data-quality checks, upsert/merge semantics, and metric versioning. Part 1: Sketch a DAG from raw events to curated snapshots to metrics. Identify tasks, dependencies, partitioning, clustering, and keys. Part 2: Write SQL that merges event-level changes into a monthly snapshot table.…