Medallion Architecture: Bronze, Silver, Gold

A layered pattern (popularized by the lakehouse architecture) for organizing data as it moves from raw capture to business-ready:

• Bronze — raw ingestion. Data as it arrived (Kafka, Kinesis, CSV/JSON files, a data lake), largely untouched.
• Silver — filtered, cleaned, augmented. Deduplicated, typed, joined with reference data — usable but not yet aggregated for a specific business question.
• Gold — business-level aggregates. The layer that actually feeds streaming analytics, AI, and reporting; this is typically where a data product's output ports read from.

Why a DPM needs this: it's the shared vocabulary for talking to data engineers about where a metric or feature actually lives, and for reasoning about root cause — a broken metric is often a Silver-layer transform problem, not a Gold-layer aggregation problem.