dbt Data Transformation Framework

dbt is a data transformation framework that lets analytics engineers write modular, tested, and version-controlled data pipelines in SQL, applying software engineering practices like code review, testing, and documentation to analytics workflows. The framework operates directly inside data warehouses including Snowflake, BigQuery, Databricks, and Redshift, compiling SQL-based model definitions into optimized queries that create materialized views, tables, incremental loads, and ephemeral computations. The Fusion engine powers local development with native SQL comprehension, catching validation errors before code reaches the warehouse and providing column-level lineage tracking through an interactive DAG that traces how fields flow and transform across models. Refactoring operations are automated: renaming a model or column propagates updates to every downstream reference, with a preview of proposed changes before commit. Multi-dialect compilation keeps transformation logic portable across data platforms, so teams can migrate warehouses without rewriting models. The testing framework runs data quality checks including not-null, unique, accepted values, and relationships constraints, with test failures halting pipeline execution before bad data reaches dashboards. IDE extensions integrate with Cursor, Claude Code, Windsurf, and VS Code, providing a native local development experience. A semantic layer defines metric dimensions centrally so that every downstream consumer, from BI tools to AI agents, computes the same numbers. The community exceeds one hundred thousand active members with eighty thousand teams using the platform weekly. Now merged with Fivetran, the combined company serves customers including Nasdaq, Toyota, Affirm, Siemens, and Sweetgreen.