Write the pipeline spec, drive an agent to build the DAG, and harden it with the retries and idempotency it won't add on its own.
Part of: Data Engineering with AI