PMLE Collaborating to manage data and models Practice Question

Which THREE actions are best practices for managing ML models in production on Google Cloud? (Choose 3)

⚠ Common exam trap

Google Cloud often tests the misconception that manual hyperparameter tuning is acceptable for production, when in fact automation (e.g., Vertex AI Vizier) is the recommended practice to ensure reproducibility and efficiency.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Monitor model performance and data drift continuously.

Option B is correct because continuous monitoring of model performance and data drift is essential in production to detect degradation and trigger retraining before business impact occurs. Option C is correct because a central model registry (such as Vertex AI Model Registry) provides governance, lineage, and controlled promotion of models across environments. Option D is correct because versioning model artifacts and training datasets ensures reproducibility, traceability, and rollback capability for every deployed model. Option A is not a best practice because manual hyperparameter tuning does not scale and should be automated with tools like Vertex AI Vizier. Option E is not a best practice because retaining all raw training data indefinitely increases cost and compliance risk; retention should follow defined policies and lifecycle rules.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Manually tune hyperparameters for each retraining run.
Why it's wrong here
Manual hyperparameter tuning for each retraining run violates the MLOps principle of automation, as Google Cloud’s Vertex AI provides automated hyperparameter tuning jobs that systematically search the parameter space. This option is tempting because manual tuning offers fine-grained control during initial model development, and would be correct in a small-scale, non-production scenario where compute cost and reproducibility are not primary concerns.
✓
Monitor model performance and data drift continuously.
Why this is correct
Continuous monitoring of model performance and data drift detects degradation after deployment, satisfying the production oversight requirement. Unlike static validation, it tracks live inference distributions against training baselines, triggering retraining when drift exceeds thresholds. This directly addresses the stem's demand for ongoing operational management of deployed ML models.
✓
Use a central model registry for model governance.
Why this is correct
A central model registry provides versioned lineage, stage transitions and approval gates for every artefact, directly satisfying the governance and reproducibility constraints of production ML. It decouples deployment from training, letting teams promote validated models while retaining audit trails — the mechanism the scenario demands for controlled, traceable releases.
✓
Version all model artifacts and training datasets.
Why this is correct
Versioning both model artefacts and training datasets creates an immutable lineage record, letting you reproduce any deployed model exactly and roll back to a prior artefact when performance degrades. This directly satisfies the stem's production-management constraint, since traceability between a prediction and the exact data and weights that produced it is otherwise unattainable.
✗
Store all raw training data indefinitely for auditability.
Why it's wrong here
Retaining all raw training data indefinitely conflicts with data-minimisation and lifecycle requirements, and storage cost grows unbounded. It tempts as an auditability safeguard, and would be correct only where regulation mandates immutable retention of specific datasets, not as a blanket production practice.

Go deeper

PMLE Collaborating to manage data and models Practice Question

Why each option matters

Option-by-option breakdown

Related to this question