CERTIFIED-MACHINE-LEARNING-PROFESSIONAL Question #19: Real Exam Question with Answer & Explanation

Question

In a continuous integration, continuous deployment (CI/CD) process for machine learning pipelines, which of the following events commonly triggers the execution of automated testing?

Options

• AThe launch of a new cost-efficient SQL endpoint
• BCI/CD pipelines are not needed for machine learning pipelines
• CThe arrival of a new feature table in the Feature Store
• DThe launch of a new cost-efficient job cluster
• EThe arrival of a new model version in the MLflow Model Registry

Explanation

Launching a new cost-efficient job cluster (D) is a common trigger in Databricks-centric ML CI/CD pipelines. Job clusters are ephemeral - they spin up on demand, run a specific job (including automated tests), and terminate immediately after. This makes them the natural execution environment and trigger point for CI/CD testing: when a pipeline stage launches a new job cluster, it signals the system to run tests in a fresh, isolated, reproducible environment.

Why the distractors are wrong:

• A (SQL endpoint): SQL compute endpoints serve analytics queries - they have no role in triggering test execution in an ML pipeline.
• B (CI/CD not needed): False. CI/CD is especially important for ML pipelines, which involve code, data, and model artifacts all changing over time.
• C (Feature Store arrival): A new feature table may trigger a retraining workflow, but it is not a standard trigger for running automated tests in CI/CD.
• E (MLflow Model Registry): A new model version arriving in the registry typically triggers deployment or validation stages, not the initial automated testing phase.

Memory tip: Think of job clusters as the "clean room" of ML CI/CD - every test runs in a fresh one. "New cluster launch = new test run" is the key association. SQL endpoints and Feature Store events are data concerns, not pipeline execution triggers.

No community discussion yet for this question.