Google Professional Machine Learning Engineer Question 382
Select 2Google Cloud PlatformYou are working on deploying a machine learning model to production using Google Cloud. The model requires preprocessing of input data, training, and evaluation. You want to automate the workflow using Google Cloud tools while optimizing compute resources and ensuring the workflow triggers on code changes. Which combination of components and triggers should you use?
- A
Use Cloud Build to orchestrate the workflow and trigger the pipeline on changes to a Git repository.
- B
Use Vertex AI Pipelines to design the workflow and trigger it manually.
- C
Use Cloud Functions to preprocess data in real-time and trigger training jobs on Vertex AI.
- D
Use Cloud Build to execute training jobs and trigger the workflow on changes to a Cloud Storage bucket.
- E
Use Vertex AI Pipelines to design the workflow and Cloud Build triggers to start the pipeline on code changes.
Show answer and explanation
Correct answers: A, E
Explanation
To automate the machine learning workflow, Vertex AI Pipelines should be used to design and manage the workflow, as it is specifically tailored for ML processes. Cloud Build triggers are ideal for initiating workflows based on code changes in a Git repository, ensuring the pipeline stays up-to-date with the latest code. Combining these tools optimizes compute resources and meets the requirements for CI/CD integration.
- A. Correct.
Correct. Cloud Build is a suitable choice for orchestrating workflows and can be configured with triggers to start the pipeline on changes to a Git repository.
- B. Incorrect.
Incorrect. While Vertex AI Pipelines is the best tool for designing workflows, triggering it manually does not meet the requirement to trigger on code changes.
- C. Incorrect.
Incorrect. Cloud Functions is not an optimal choice for orchestrating training pipelines and lacks the direct integration needed for large-scale ML workflows.
- D. Incorrect.
Incorrect. Cloud Build is not suitable for executing large-scale ML training jobs, as it is primarily intended for CI/CD tasks like building and deploying code.
- E. Correct.
Correct. Vertex AI Pipelines is the ideal choice for designing ML workflows, and using Cloud Build triggers ensures the pipeline executes on code changes, satisfying the requirements.