nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #245

You work for a retail company that is using a regression model built with BigQuery ML to predict product sales. This model is being used to serve online predictions. Recently you developed a new…

The correct answer is C. Import the new model to the same Vertex AI Model Registry as the existing model. Deploy each. To minimize impact while deploying a new model version for monitoring, import both models as distinct versions to the same Vertex AI Model Registry and deploy them to the same Vertex AI endpoint, allocating traffic using a controlled rollout.

Submitted by brentm· Apr 18, 2026ML pipeline operationalization

Question

You work for a retail company that is using a regression model built with BigQuery ML to predict product sales. This model is being used to serve online predictions. Recently you developed a new version of the model that uses a different architecture (custom model). Initial analysis revealed that both models are performing as expected. You want to deploy the new version of the model to production and monitor the performance over the next two months. You need to minimize the impact to the existing and future model users. How should you deploy the model?

Options

  • AImport the new model to the same Vertex AI Model Registry as a different version of the existing
  • BImport the new model to the same Vertex AI Model Registry as the existing model. Deploy the
  • CImport the new model to the same Vertex AI Model Registry as the existing model. Deploy each
  • DDeploy the new model to a separate Vertex AI endpoint. Create a Cloud Run service that routes

How the community answered

(52 responses)
  • A
    10% (5)
  • B
    4% (2)
  • C
    71% (37)
  • D
    15% (8)

Why each option

To minimize impact while deploying a new model version for monitoring, import both models as distinct versions to the same Vertex AI Model Registry and deploy them to the *same* Vertex AI endpoint, allocating traffic using a controlled rollout.

AImport the new model to the same Vertex AI Model Registry as a different version of the existing

While importing as a different version is correct, it doesn't specify how to *deploy* and *monitor* without impact, which is crucial for a controlled rollout.

BImport the new model to the same Vertex AI Model Registry as the existing model. Deploy the

Deploying *only* the new model and replacing the existing one would have a high impact and prevent direct comparison of performance over time.

CImport the new model to the same Vertex AI Model Registry as the existing model. Deploy eachCorrect

Importing both models as separate versions under the same Vertex AI Model in the Model Registry is correct for tracking. Deploying both the existing and new model versions to the *same* Vertex AI endpoint and using traffic splits minimizes impact by allowing a controlled rollout and direct performance comparison, which is essential for monitoring.

DDeploy the new model to a separate Vertex AI endpoint. Create a Cloud Run service that routes

Deploying to a *separate* Vertex AI endpoint and using Cloud Run to route traffic introduces unnecessary complexity and an additional layer of management compared to leveraging Vertex AI's native traffic splitting capabilities on a single endpoint.

Concept tested: Vertex AI model deployment, traffic splitting, A/B testing

Source: https://cloud.google.com/vertex-ai/docs/predictions/deploy-model-api#endpoint-traffic-split

Topics

#Vertex AI#Model Deployment#Traffic Splitting#Canary Deployment

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice