Building dbt Pipelines with Snowflake Cortex Code: A Hands-On Guide

A hands-on walkthrough of using Snowflake Cortex Code CLI to build a complete dbt project against live Snowflake data in a single session. Cortex Code profiled raw tables with intentional data quality issues (duplicate IDs, NULL fields, mixed date formats, VARCHAR-stored numbers), then generated defensive staging models with deduplication and safe casting, a marts layer with business aggregations, and 17 dbt tests — all passing. The full workflow including schema creation, model generation, and test execution completed in under 8 minutes. Key patterns covered include ROW_NUMBER deduplication, TRY_TO_DATE for resilient parsing, COALESCE defaults, and PAT-based authentication via environment variables.