MLA-C02 practice question: SageMaker Pipelines and CI/CD (1 of 2)
Deployment and Orchestration of ML and AI Workflows A free question from our original MLA-C02 practice exam. Try it, then open the answer to see why each option is right or wrong.
A dental chain retrains its patient no-show model each week. The team wants a SageMaker Pipelines workflow that trains the model, evaluates it on a holdout set, and registers the model only when its AUC is at or above the value the production model achieved.
A model that fails the check must not be registered at all, and a model that passes must be registered as pending manual approval so that a reviewer signs it off before production.
Which pipeline design meets these requirements?
Choose one answer.
AA. Train and evaluate the model, use a condition step to compare its AUC with the threshold, and on success register the version as PendingManualApproval.
BB. Train, evaluate and register the model in a single script step, and have the script exit with an error code when the AUC is too low.
CC. Register every trained model, and add a separate approval step that deletes the models whose AUC is too low after registration.
DD. Trigger the pipeline and then have an engineer check the AUC by hand before deciding whether to register the model.
Show the answer and explanation
Answer: A
A. Right: Correct. A condition step routes the pipeline run on the evaluation metric, so the model step that registers the version runs only when the check passes. Registering it with PendingManualApproval keeps a reviewer in the loop before production. Learn more: Add a step to a pipeline (opens in a new tab)
B. Wrong: A script step that fails stops the pipeline, but the check is hidden inside custom code rather than a pipeline control flow step. The stem asks for a pipeline design that gates registration explicitly.
C. Wrong: Registering weak models first puts them in the registry, which the stem forbids. Deleting them afterward is cleanup, not a gate.
D. Wrong: A manual check does not automate the gate the team wants, and the stem asks for registration to depend on the metric. The option relies on people to apply the rule.
Explanation
A condition step in SageMaker Pipelines compares the evaluation metric with the threshold, and only a passing run reaches the step that registers the model version as pending manual approval. Registering first, hiding the gate in a script or relying on a manual check does not enforce the rule in the pipeline.
Want more on SageMaker Pipelines and CI/CD? The 5 full-length practice sets have 9 more SageMaker Pipelines and CI/CD questions, 325 in all, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets →