Google Professional Machine Learning Engineer Question 423
Select 4Google Cloud PlatformYour team is responsible for deploying a new machine learning model to production on Google Cloud. You want to implement a CI/CD pipeline to automate the deployment process while ensuring that only fully validated models are deployed. Which combination of actions should you take to achieve this goal?
- A
Use Cloud Build to automate the pipeline and include validation tests as pipeline steps.
- B
Store the model artifacts in a Cloud Storage bucket with versioning enabled.
- C
Deploy the model directly to production from a local machine after running manual tests.
- D
Use Vertex AI Model Registry to manage model versions and set up triggers for deployment.
- E
Implement a canary deployment strategy using Cloud Run to test the model with a subset of traffic.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
To implement a CI/CD pipeline for deploying machine learning models, you need to automate the process, ensure proper version control, and validate the model before deployment. Cloud Build can automate the pipeline, Cloud Storage with versioning ensures model traceability, Vertex AI Model Registry offers version management, and a canary deployment strategy helps mitigate risks during production rollout. Avoid manual deployments from local machines, as they are prone to errors and do not align with CI/CD practices.
- A. Correct.
Correct: Cloud Build is a fully managed CI/CD tool that can automate pipelines, including running validation tests as part of the deployment process.
- B. Correct.
Correct: Storing model artifacts in a Cloud Storage bucket with versioning ensures traceability and provides a mechanism to roll back to previous versions if needed.
- C. Incorrect.
Incorrect: Deploying directly from a local machine bypasses automation and validation steps, which is against CI/CD best practices.
- D. Correct.
Correct: Vertex AI Model Registry helps manage model versions and can trigger automated deployment pipelines, aligning with CI/CD principles.
- E. Correct.
Correct: Canary deployments allow you to test a new model with a subset of production traffic, providing a controlled rollout mechanism.