dbt-polars

dbt-polars integrates the flexibility and performance of Polars with the data governance and lineage capabilities of dbt.

Key features of dbt-polars:

• Run anywhere: Polars runs on any compute platform, so you’re not locked into an expensive data warehouse.
• Flexible: Available catalogs let you connect to local storage, Iceberg catalogs, Databricks, and more.
• Extensible: Create a custom catalog to fit your unique storage and warehouse needs.
• First-class Python models: Run Python models with the same performance and capabilities as SQL models.
• Mix Python and SQL: Choose the language that best fits the transformation.
• dbt-native: Bring lineage, documentation, and data testing into your workflow.
• Ingestion: Build efficient, incremental API ingestion with Python models and keep the entire pipeline visible in the lineage graph.

Installation

pip install dbt-polars

Extra dependencies are required for cloud and Iceberg backends:

pip install 'dbt-polars[azure]'       # Azure Blob Storage
pip install 'dbt-polars[databricks]'  # Azure Blob Storage
pip install 'dbt-polars[fabric]'      # Microsoft Fabric Lakehouse
pip install 'dbt-polars[snowflake]'   # Snowflake (Iceberg tables)
pip install 'dbt-polars[iceberg]'     # Apache Iceberg (SQLite, REST, Databricks, …)

Catalogs

dbt-polars uses catalogs to define where data is stored. Each catalog maps to a storage backend. You can configure multiple catalogs in a single profile and reference them from your models.

Quick start

See the Getting started guide.

A minimal profiles.yml for local development is:

my_project:
  target: dev
  outputs:
    dev:
      type: polars
      schema: dev

      catalog_type: local
      root: ./data

Documentation

• Getting started
• Catalog overview and profile reference
• Local catalog
• Azure catalog
• Iceberg catalog
• Databricks catalog
• Fabric catalog
• Snowflake catalog
• Custom catalogs
• Python models and tests