dbt Examples — Warehouse-native Analytics Engineering

Production-grade dbt patterns across four focused sub-projects: medallion modeling, data quality testing, incremental strategies, and performance & cost optimization.

⚠️ Educational & academic use only

This repository is published exclusively for educational, academic, and demonstrative purposes.

• All data is fictional — generated programmatically with deterministic seeds. No client data, no real company data, no production data, no PII.
• The code is teaching material, not a consulting deliverable, not a client implementation, not production-tuned configuration.
• No warranty — distributed under Apache 2.0 AS IS. Anyone using this code in a real environment does so at their own risk.
• No confidential information of clients, partners, or third parties is included.

Full legal terms: see DISCLAIMER.md and LICENSE.

---

Why this repo exists

dbt is the de facto standard for transformation inside the modern warehouse (Snowflake, BigQuery, Redshift, Databricks SQL, DuckDB). These examples mirror what a real analytics engineering team ships to production — not toy notebooks.

Each sub-project is a self-contained dbt project that runs out of the box against DuckDB (zero credentials needed), and is documented for porting to Snowflake / BigQuery / Redshift / Databricks.

---

Sub-projects

Project	Topic	What it shows
dbt_medallion	Staging → Intermediate → Marts	Sources, contracts, snapshots, exposures, macros, full warehouse-native medallion
dbt_testing	Data quality that doesn't ping at 1 AM	Generic + singular tests, dbt-expectations, unit tests (dbt 1.8+), Elementary patterns
dbt_incremental	Incremental strategies side-by-side	merge vs insert_overwrite vs append vs microbatch (1.9), backfill safety, late-arriving data
dbt_performance	Cost & performance optimization	Materialization choice, clustering, query tagging, warehouse spend monitoring with metadata

---

Quick start (zero credentials)

# Requires Python 3.10+
pip install dbt-duckdb

cd dbt_medallion
dbt deps
dbt seed
dbt build
dbt docs generate && dbt docs serve

Each sub-project ships with a profiles.yml.example pointing at a local DuckDB file. To run against Snowflake / BigQuery / Redshift, copy that file to ~/.dbt/profiles.yml and swap the adapter block — instructions are inside each sub-project's README.

---

Tested adapter matrix

Adapter	Status	Notes
DuckDB	✅ Default for examples	Zero credentials, runs in CI
Snowflake	✅ Documented	Performance examples (dbt_performance) include Snowflake-specific patterns
BigQuery	✅ Documented	Incremental microbatch examples include partition/cluster recommendations
Redshift	✅ Documented	merge strategy notes for Redshift's limitations
Databricks SQL	✅ Compatible	Pairs naturally with the databricks-examples companion repo

---

Repository conventions

• One project = one topic. No cross-imports between sub-projects.
• Seeds are deterministic. Fake data is generated with fixed seeds so test results are reproducible.
• Every model has a description. If dbt docs generate shows blanks, that's a bug.
• Every public model has at least one not_null and one unique test on its primary key. Period.
• No dbt run in CI — only dbt build. Because tests are not optional.

---

How dbt is operated in production

Curious how a real team runs dbt in production (Kubernetes, GitHub Actions, approval gates)? See docs/DEPLOYMENT_PATTERN.md — explanatory, not executable.

---

License

Apache 2.0 — see LICENSE.

---

About

Built by Napoli Data — Data & AI Engineering. We publish in the open what others charge for. If this is what we give away, imagine what we ship under contract.

Companion repo (Lakehouse / PySpark): databricks-examples