PMLE Practice Question: Wants to implement continuous training for a…

An organization wants to implement continuous training for a model that serves predictions via Vertex AI Endpoints. Which approach best automates the retrain-deploy cycle?

⚠ Common exam trap

Google Cloud often tests the distinction between partial automation (e.g., only retraining or only deploying) and full end-to-end automation; the trap here is that candidates may choose an option that automates only one part of the cycle (like retraining with Cloud Composer or auto-deployment with Model Registry) and miss that the question requires both retraining and deployment to be automated in a single, orchestrated workflow.

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

✓

Schedule a Vertex AI Pipeline to retrain and conditionally deploy

Vertex AI Pipelines can be scheduled to run a retraining workflow and include a conditional step that deploys the new model to the endpoint only if it passes validation (e.g., evaluation metrics meet a threshold). This fully automates the retrain-deploy cycle without manual intervention, leveraging the pipeline's orchestration capabilities.

Answer analysis

Option-by-option breakdown

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

✓
Schedule a Vertex AI Pipeline to retrain and conditionally deploy
Why this is correct
A scheduled Vertex AI Pipeline can retrain on a defined cadence and use a condition to deploy only when evaluation metrics pass, directly automating the retrain-deploy cycle for models served through Vertex AI Endpoints without manual intervention.
✗
Use Vertex AI Model Registry to auto-deploy on new model upload
Why it's wrong here
The Model Registry stores and versions models but does not itself trigger training or automatically deploy to an endpoint on upload. It is tempting because registry-centric promotion feels like the natural automation hub, which would be correct if the question asked how to track model lineage and gate versions for approval before release.
✗
Manually retrain and deploy monthly
Why it's wrong here
Monthly manual retraining introduces human scheduling and intervention, so the retrain-deploy cycle is not automated and drifts between runs. It is tempting because a fixed cadence is easy to document and audit, which would suit a stable model where regulations require predictable, human-approved release windows rather than continuous automation.
✗
Use Cloud Composer to schedule retraining only
Why it's wrong here
Cloud Composer can schedule the retraining job, but scheduling alone leaves deployment to the endpoint unhandled, so the retrain-deploy cycle stays incomplete. It is tempting because Composer orchestrates complex DAGs well, which would be correct if the task required coordinating multi-step data preparation dependencies rather than end-to-end automation.
✗
Use a Cloud Function to retrain the model and update the endpoint
Why it's wrong here
A Cloud Function is a short-lived event handler, not a training orchestrator, so it cannot run the pipeline, evaluate the model, or manage endpoint rollout. It is tempting because serverless functions trigger cheaply on events, which would be correct for lightweight tasks such as firing a notification when new training data lands in a bucket.

Go deeper

PMLE Practice Question: Wants to implement continuous training for a…

Why each option matters

Option-by-option breakdown

Related to this question