Data Processing & Transformations

The raw landing layer: data is ingested append-only, schema-on-read, preserving full source fidelity and history.

So it preserves a complete, immutable history of source data, enabling reprocessing/replay of downstream layers from raw.

Cleaned, validated, deduplicated, conformed and joined data forming the 'enterprise view' used across many use cases.

Because it provides clean, conformed, reusable data that many downstream teams and Gold marts can build on.

Business-level aggregates and project-specific data marts, ready for BI dashboards and ML feature consumption.

The Gold layer, because it contains curated, aggregated, business-ready data.