Google Professional Machine Learning Engineer Question 311
Select 3Google Cloud PlatformYou are working on a machine learning project where multiple teams collaborate on developing and deploying models. To ensure proper versioning, tracking, and management of the models produced, you decide to organize a model registry using Google Cloud tools. Which practices should you implement to effectively set up and maintain the model registry?
- A
Use Vertex AI Model Registry to version and track metadata for all models.
- B
Ensure each model version is tagged with relevant metadata such as performance metrics, deployment stage, and owner.
- C
Store all model artifacts in a local file system to minimize costs.
- D
Implement a process to archive deprecated models while retaining their metadata for future audits.
- E
Allow only the development team to access and modify the model registry.
Show answer and explanation
Correct answers: A, B, D
Explanation
To effectively organize a model registry in Google Cloud, you should leverage tools like Vertex AI Model Registry, which provide robust versioning and metadata management capabilities. Tagging models with relevant metadata helps maintain clarity and traceability across teams. Archiving deprecated models ensures auditability and compliance while avoiding unnecessary clutter. However, relying on local storage or overly restrictive access policies can limit collaboration and scalability, which are essential in modern machine learning workflows.
- A. Correct.
Vertex AI Model Registry is purpose-built for managing and versioning machine learning models in Google Cloud. It enables tracking metadata and managing model lifecycles effectively.
- B. Correct.
Tagging models with metadata such as performance metrics, deployment stage, and ownership is critical for tracking, versioning, and understanding the context of each model.
- C. Incorrect.
Storing model artifacts in a local file system is not recommended for collaborative machine learning workflows as it lacks scalability, versioning support, and integration with cloud services.
- D. Correct.
Archiving deprecated models while retaining metadata ensures compliance, auditability, and the ability to revisit past models if needed.
- E. Incorrect.
Restricting access to only the development team may hinder collaboration between operations, data science, and DevOps teams, which are often involved in model deployment and monitoring.