dbt-expert

Organize dbt models into a layered architecture where staging models clean raw source data, intermediate models perform business logic transformations, and mart models serve final analytical tables. This structure enforces data lineage, improves maintainability, and enables consistent testing and documentation across your dbt project.

Define schema tests in YAML files alongside your dbt models to validate data quality—checking for unique values, non-null constraints, and referential integrity. Tests run during `dbt test` execution and catch data issues early, ensuring only validated data flows downstream.

Add semantic YAML descriptions directly in your model files and configure dbt_project.yml to enforce naming conventions and documentation standards. This creates discoverable, maintainable documentation that propagates to your dbt docs site and ensures all team members follow the same structure.

Use `source()` to reference raw tables in your data warehouse and `ref()` to reference other dbt models. This distinction establishes clear data lineage: sources mark entry points from external systems, and refs connect your transformation DAG, enabling dbt to understand dependencies and rebuild models correctly.

Define reusable SQL logic in macros and store static reference data in seed files (CSV format). Configure both in dbt_project.yml to ensure your entire project—transformations, tests, and static data—runs consistently across environments and team members.

Yes. Best practices cover the full dbt workflow: standardized source definitions, layered model patterns, comprehensive schema and data tests, and semantic documentation. This approach scales from source ingestion through transformation to analytics-ready marts while maintaining data quality and project clarity.