dbt Data Modeling

dbt's docs cover features. This skill covers layout and discipline — the parts that decide whether your warehouse stays understandable past the first 50 models.

When to use

• New dbt project (Snowflake / BigQuery / Redshift / Postgres / DuckDB).
• Refactoring a project that's grown without layering.
• Setting up CI for dbt for the first time.

When NOT to use

• Pure transformation jobs that don't need lineage (one-off Python scripts).
• Streaming pipelines — dbt is batch-first; use Materialize or RisingWave.

The layering (the only thing that matters)