Google Professional Machine Learning Engineer Question 146
Single answerGoogle Cloud PlatformYou are a Machine Learning Engineer tasked with building a Jupyter notebook environment for training and experimenting with machine learning models on Google Cloud. Your workflow involves preprocessing a large dataset stored in BigQuery, training a TensorFlow model using GPUs, and deploying the model on Vertex AI. Which Jupyter backend should you choose for this workflow?
- A
AI Platform Notebooks
- B
Vertex AI Workbench (Managed Notebooks)
- C
Dataproc Cluster with JupyterHub
- D
Cloud Datalab
Show answer and explanation
Correct answer: B
Explanation
Vertex AI Workbench (Managed Notebooks) is the most appropriate Jupyter backend for this scenario because it provides an end-to-end environment for machine learning tasks. It includes built-in integration with BigQuery for data preprocessing, TensorFlow and GPU support for training, and Vertex AI for deployment. Other options either lack the required features, are legacy tools, or are not optimized for the given workflow.
- A. Incorrect.
AI Platform Notebooks is a legacy option for running Jupyter environments on Google Cloud. While it supports some ML workflows, it lacks the seamless integration and managed features provided by Vertex AI Workbench.
- B. Correct.
Vertex AI Workbench (Managed Notebooks) is specifically designed for machine learning workflows. It offers tight integration with BigQuery, TensorFlow, GPUs, and Vertex AI, making it the most suitable choice for the described workflow.
- C. Incorrect.
Dataproc Cluster with JupyterHub is primarily used for distributed data processing and analysis with Spark or Hadoop. It is not optimized for TensorFlow model training with GPUs or deploying models on Vertex AI.
- D. Incorrect.
Cloud Datalab is a deprecated tool that is no longer actively supported or recommended for new machine learning workflows.