Google Professional Machine Learning Engineer Question 398
Select 3Google Cloud PlatformYou are designing a machine learning pipeline for a model that processes real-time customer interactions and retrains the model daily using updated data. You decide to use Google Cloud Composer to orchestrate the pipeline. Which of the following tasks can be effectively performed using Cloud Composer in this scenario?
- A
Scheduling daily retraining jobs for the machine learning model
- B
Monitoring the performance of the trained machine learning model in production
- C
Triggering a Dataflow job to preprocess incoming data for training
- D
Storing the trained model artifacts in Cloud Storage
- E
Managing dependencies between pipeline steps such as training, evaluation, and deployment
Show answer and explanation
Correct answers: A, C, E
Explanation
Cloud Composer is a managed workflow orchestration service built on Apache Airflow. It is designed to automate and manage complex workflows by defining tasks and their dependencies in DAGs. In this scenario, tasks like scheduling retraining jobs, triggering external services (e.g., Dataflow), and managing dependencies between pipeline steps are well-suited for Cloud Composer. However, tasks like monitoring model performance or directly storing artifacts are outside its scope and are better handled by other Google Cloud services.
- A. Correct.
Scheduling daily retraining jobs is a key feature of Cloud Composer. It allows you to schedule and automate workflows, making it suitable for orchestrating periodic retraining tasks.
- B. Incorrect.
Monitoring the performance of the model in production is not a direct feature of Cloud Composer. This task is better handled using monitoring tools like Cloud Monitoring or AI Platform Pipelines.
- C. Correct.
Cloud Composer can trigger external services, such as a Dataflow job, as part of a pipeline. This makes it capable of orchestrating preprocessing steps in the workflow.
- D. Incorrect.
While Cloud Composer can trigger a task to store model artifacts in Cloud Storage, it is not directly responsible for storing artifacts. This task is typically handled by the model training or deployment job itself.
- E. Correct.
Managing dependencies between pipeline steps is one of the primary uses of Cloud Composer. By defining DAGs (Directed Acyclic Graphs), you can ensure that tasks such as training, evaluation, and deployment are executed in the correct order.