A city wants notebooks and pipeline definitions reviewed before production deploy. What SDLC practice should they apply?

A city that wants notebooks and pipeline definitions reviewed before production deploy should store Databricks notebooks and pipeline definitions in Git and require code review before production deploy. Git provides history, blame, rollback, and a pull-request gate; code review catches logic, security, and cost issues before they hit municipal production. Treating notebooks as uncontrolled workspace-only artifacts is a common anti-pattern the exam expects you to reject.

Storing Databricks notebooks and pipeline definitions in DBFS and requiring code review before production deploy fails because DBFS is runtime file storage, not version control—it lacks Git history, blame, branching, and a pull-request gate, so there is no durable reviewed merge to enforce before production deploy. Storing Databricks notebooks and pipeline definitions in Git and requiring data profile approval before production deploy fails because data profile approval validates data outputs and quality, not code changes, and it does not provide peer code review for logic, security, and cost before production deploy. Storing Databricks notebook run outputs and pipeline run logs in Git and requiring code review before production deploy fails because Git should version source notebooks and pipeline definitions, not generated outputs and logs, which bloat the repository and do not allow review of code changes before production deploy.

Exam caveat: Git + review is the SDLC baseline; Asset Bundles and CI/CD often package the deploy step, but the question focuses on version control and review before prod. Repos should include notebooks, pipeline JSON/YAML, and job definitions as appropriate—not screenshots. Branch protections on main reinforce the review gate.

Operational check: confirm the civic Databricks assets live in a remote Git repo, that production deploy requires a reviewed merge (PR), and that storing notebooks and pipeline definitions in DBFS is not accepted as a substitute for Git versioning with a reviewed merge before production.