Capstone: apply, push, run, query

You built every layer. The capstone wires them together. Terraform applies. Push to main triggers Cloud Build. Cloud Build deploys Cloud Run. Airflow triggers /run_transformation. dbt builds the marts. BigQuery returns the latest row. End to end.

You now own a working GCP analytics stack: GCS, BigQuery, dbt, Cloud Run, Airflow, Cloud Build, Terraform. The same shape with engine swaps gives you Snowflake plus dbt, Redshift plus dbt, or Postgres plus dbt. The pattern transfers; the engines are the easy part.