Google Professional Machine Learning Engineer Question 389
Single answerGoogle Cloud PlatformYou are designing a machine learning workflow for a model that will process large-scale time-series data. The process involves data preprocessing, feature extraction, model training, and evaluation. You need to choose an orchestration framework that integrates seamlessly with Google Cloud services, supports reproducibility, and allows you to monitor and manage the pipeline. Which orchestration framework should you choose?
- A
Kubeflow Pipelines
- B
Apache Airflow
- C
Vertex AI Pipelines
- D
TensorFlow Extended (TFX)
Show answer and explanation
Correct answer: C
Explanation
Vertex AI Pipelines is the best choice in this scenario because it is a fully managed service on Google Cloud that is specifically designed for orchestrating machine learning pipelines. It provides seamless integration with other Google Cloud services like BigQuery, Cloud Storage, and Vertex AI, and it supports reproducibility and monitoring for end-to-end workflows. While other tools like Kubeflow Pipelines and Apache Airflow have their strengths, they may require more setup and do not offer the same level of integration or specialized ML features as Vertex AI Pipelines.
- A. Incorrect.
Kubeflow Pipelines is an orchestration tool that is well-suited for Kubernetes-based environments and supports machine learning workflows, but it may require more setup and maintenance compared to Vertex AI Pipelines when used on Google Cloud.
- B. Incorrect.
Apache Airflow is a general-purpose orchestration tool, but it is not specifically optimized for machine learning workflows and does not integrate as seamlessly with Google Cloud ML services as Vertex AI Pipelines.
- C. Correct.
Vertex AI Pipelines is a managed service on Google Cloud that is specifically designed for building and managing reproducible machine learning pipelines. It integrates seamlessly with other Google Cloud services and provides built-in monitoring and management features.
- D. Incorrect.
TensorFlow Extended (TFX) is a framework for building ML pipelines, but it is not an orchestration tool itself. It can be used in conjunction with orchestration tools like Kubeflow Pipelines or Vertex AI Pipelines.