nerdexam
Google

PROFESSIONAL-MACHINE-LEARNING-ENGINEER · Question #270

You have developed an application that uses a chain of multiple scikit-learn models to predict the optimal price for your company's products. The workflow logic is shown in the diagram. Members of…

The correct answer is C. Expose each individual model as an endpoint in Vertex AI Endpoints. Use Cloud Run to. To deploy a chained scikit-learn model workflow with individual model version control, zero-scaling, and minimal compute/effort, deploy each model to Vertex AI Endpoints and orchestrate with Cloud Run.

Submitted by kim_seoul· Apr 18, 2026ML pipeline operationalization

Question

You have developed an application that uses a chain of multiple scikit-learn models to predict the optimal price for your company's products. The workflow logic is shown in the diagram. Members of your team use the individual models in other solution workflows. You want to deploy this workflow while ensuring version control for each individual model and the overall workflow. Your application needs to be able to scale down to zero. You want to minimize the compute resource utilization and the manual effort required to manage this solution. What should you do?

Options

  • AExpose each individual model as an endpoint in Vertex AI Endpoints. Create a custom container
  • BCreate a custom container endpoint for the workflow that loads each model's individual files Track
  • CExpose each individual model as an endpoint in Vertex AI Endpoints. Use Cloud Run to
  • DLoad each model's individual files into Cloud Run. Use Cloud Run to orchestrate the workflow.

How the community answered

(66 responses)
  • A
    12% (8)
  • B
    6% (4)
  • C
    50% (33)
  • D
    32% (21)

Why each option

To deploy a chained scikit-learn model workflow with individual model version control, zero-scaling, and minimal compute/effort, deploy each model to Vertex AI Endpoints and orchestrate with Cloud Run.

AExpose each individual model as an endpoint in Vertex AI Endpoints. Create a custom container

While individual endpoints are good, using a custom container on Vertex AI Endpoints for the *workflow* itself might not scale down to zero as effectively as Cloud Run, and duplicates orchestration logic in the container.

BCreate a custom container endpoint for the workflow that loads each model's individual files Track

Creating a single custom container endpoint for the entire workflow makes individual model versioning and reuse difficult for other workflows and hinders the ability to scale different parts of the chain independently.

CExpose each individual model as an endpoint in Vertex AI Endpoints. Use Cloud Run toCorrect

Deploying each individual scikit-learn model to its own Vertex AI Endpoint enables independent version control and management for each component. Cloud Run, being serverless, can then orchestrate calls to these individual endpoints, scaling down to zero when idle, minimizing compute utilization, and reducing manual effort for the overall workflow.

DLoad each model's individual files into Cloud Run. Use Cloud Run to orchestrate the workflow.

Loading individual model files directly into Cloud Run for serving is possible, but Vertex AI Endpoints are purpose-built for managed model serving, offering better features for version control, monitoring, and dedicated scaling for ML models, while Cloud Run is best for orchestrating the overall flow.

Concept tested: Microservices for ML with Vertex AI and Cloud Run

Source: https://cloud.google.com/vertex-ai/docs/predictions/deploy-model-custom-container

Topics

#MLOps#Serverless ML#Vertex AI Endpoints#Cloud Run

Community Discussion

No community discussion yet for this question.

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