dbt-skill

Structure dbt projects using Medallion or Kimball architectures with staging and marts layers. This approach separates raw transformations from presentation models, ensuring maintainable data modeling and clear naming conventions across your data warehouse.

Test data quality in dbt using schema, singular, generic, and unit tests. You can enhance anomaly detection by integrating dbt-expectations and Elementary packages to validate metrics and catch edge cases before production deployment.

Set up dbt CI/CD workflows using Slim CI, GitHub Actions, or dbt Cloud jobs. These methods validate model changes through automated testing before deployment, supporting strategies like blue/green deployments for zero-downtime production releases.

Optimize dbt incremental materializations on Snowflake and BigQuery by applying specific performance tuning techniques. This includes configuring incremental strategies and monitoring warehouse costs to reduce query compute overhead.

Use dbt Mesh and model versioning when managing complex cross-project dependencies. Model contracts enforce schema guarantees between projects, while dbt Mesh allows decentralized analytics engineering teams to build interconnected data pipelines.

Configure the dbt semantic layer by defining metrics in YAML files and leveraging Jinja macros for dynamic SQL generation. This enables reusable analytics engineering logic and consistent metric definitions across your data warehouse platforms.