How Sigma manages the semantic layer with dbt and Dagster

Sigma's internal data platform team manages every semantic asset—Snowflake semantic views, Cortex Agents, Cortex Search Services, and Sigma Data Models—inside a single dbt project. In this session, we'll walk through the Dagster job that continuously deploys these assets whenever they change: building modified dbt nodes, persisting dbt state, updating the Sigma connection, and refreshing Sigma Data Models. We'll also preview an open-source contribution to OSI/Ossie: a Sigma converter that lets any team convert between semantic formats within a Dagster job, making it easier to adopt a unified, orchestrated semantic layer across tools.