Le Duy Khuong (Daniel)

Series: lakehouse-layer-spec · Part 0

Dev Productivity & Tools

Series: Lakehouse Layer Spec (7 layers)

Intro to a 7-post series specifying each Lakehouse layer: Raw, Staging, Curated, Analytics, Metadata & Governance, Query & BI, Consumption.

2026-03-171 min read

Series: lakehouse-layer-spec
  1. 0.Series: Lakehouse Layer Spec (7 layers)(this post)

This 7-post series describes each layer in a Lakehouse architecture (raw → consumption):

  1. Layer 1 — Raw: Source data, immutable, ingested from sources.
  2. Layer 2 — Staging: Cleansing, normalization, validation.
  3. Layer 3 — Curated: Business and domain enrichment, ready for BI/ML.
  4. Layer 4 — Analytics: Aggregations, pre-aggregation, KPIs.
  5. Layer 5 — Metadata & Governance: Catalog, lineage, ownership, policy.
  6. Layer 6 — Query & BI: SQL engine, views, BI integration.
  7. Layer 7 — Consumption: Dashboards, APIs, ML serving.

Reference stack: Airbyte, Kafka, Spark, dbt, Iceberg/Delta, MinIO, DataHub, Trino, Superset. Start with Layer 1 — Raw.

LDK

Le Duy Khuong

AI Transformation & Digital Strategy. Writing about agentic systems, engineering leadership, and building in public.