Google Professional Machine Learning Engineer Question 433
Select 3Google Cloud PlatformYou are a Machine Learning Engineer at a retail company using Google Cloud to build and deploy a recommendation model. Your team wants to ensure that all metadata, such as dataset versions, hyperparameters, and model evaluation metrics, are tracked and auditable to comply with internal governance policies. Which combination of tools and practices should you implement?
- A
Use Vertex AI Metadata to track datasets, models, and pipelines.
- B
Store all metadata in a Google Sheets document for easy sharing.
- C
Implement Vertex AI Pipelines to automatically log metadata during pipeline execution.
- D
Use BigQuery to store and query custom metadata for advanced analysis.
- E
Enable Cloud Logging and Cloud Audit Logs to track infrastructure and API access.
Show answer and explanation
Correct answers: A, C, E
Explanation
Tracking and auditing metadata is critical for reproducibility and compliance in machine learning workflows. Vertex AI Metadata and Vertex AI Pipelines are purpose-built for managing ML metadata, while Cloud Logging and Cloud Audit Logs ensure comprehensive auditing of infrastructure and API usage. Together, these tools provide a robust solution for tracking and auditing metadata at scale.
- A. Correct.
Vertex AI Metadata is specifically designed to track and manage metadata related to datasets, models, and pipelines, making it an essential tool for ML metadata tracking.
- B. Incorrect.
While Google Sheets can be used for manual metadata tracking, it is not scalable, auditable, or suitable for large-scale ML operations.
- C. Correct.
Vertex AI Pipelines seamlessly logs metadata during pipeline execution, ensuring reproducibility and auditability of ML workflows.
- D. Incorrect.
BigQuery is powerful for querying structured data but is not specifically designed for tracking ML metadata, although it can complement metadata analysis use cases.
- E. Correct.
Cloud Logging and Cloud Audit Logs provide detailed logs of infrastructure usage and API access, which are essential for auditing and governance.