dbt Data Transformation Workflow

Explain how dbt works and its role in the modern data stack.

What Is dbt

dbt (data build tool) is a SQL-first transformation framework that lets data engineers write version-controlled, testable transformation logic using SQL and Jinja templates.

Core Concepts

Model: Each .sql file defines a SELECT statement. dbt compiles it into a CREATE TABLE AS or CREATE VIEW AS statement and runs it in the warehouse.

Materialization Types:

• view: SQL executes at query time; no data stored.
• table: Results stored as a physical table; fast queries but full rebuild each run.
• incremental: Only processes new/updated data; suitable for large tables.

Ref Function: {{ ref("model_name") }} references other models; dbt automatically resolves dependency order.

Tests: Declarative data quality checks (not_null, unique, accepted_values).

Documentation: Auto-generates lineage graphs and a data dictionary.

Layer Architecture

Staging (raw cleaning) → Intermediate (business logic) → Marts (analyst-facing final tables).