Google Professional Machine Learning Engineer Question 362
Select 3Google Cloud PlatformYour team is tasked with building an end-to-end machine learning pipeline for a retail company to predict product demand. The pipeline must handle data ingestion, preprocessing, model training, evaluation, and deployment. Which considerations are critical for designing this pipeline in Google Cloud to ensure scalability, reliability, and maintainability?
- A
Use Google Cloud Dataflow for scalable data preprocessing and transformation.
- B
Hard-code model hyperparameters directly into the training script to simplify deployment.
- C
Leverage Vertex AI Pipelines to orchestrate and automate the pipeline steps.
- D
Store raw and preprocessed data in BigQuery for efficient querying and analysis.
- E
Deploy the model directly from the local machine to minimize latency.
Show answer and explanation
Correct answers: A, C, D
Explanation
Designing an end-to-end ML pipeline on Google Cloud requires leveraging scalable and managed services to ensure the pipeline can handle large datasets, enable automation, and remain maintainable. Dataflow is ideal for scalable preprocessing, Vertex AI Pipelines simplifies orchestration, and BigQuery efficiently handles data storage and querying. Avoid hardcoding hyperparameters and deploying directly from a local machine, as these practices hinder scalability and flexibility.
- A. Correct.
Using Google Cloud Dataflow allows for scalable data preprocessing and transformation, which is essential for handling large datasets and ensuring that the pipeline can scale with the business needs.
- B. Incorrect.
Hard-coding hyperparameters into the training script is not a best practice because it reduces flexibility and makes experimentation and tuning more difficult. Instead, use parameterization or configuration files.
- C. Correct.
Vertex AI Pipelines is specifically designed for orchestrating and automating ML workflows on Google Cloud, ensuring maintainability and reducing manual intervention.
- D. Correct.
BigQuery is a highly scalable and efficient solution for storing and analyzing large datasets. It supports SQL-like queries and integrates well with other Google Cloud services, making it a good choice for raw and preprocessed data storage.
- E. Incorrect.
Deploying the model directly from a local machine is not recommended. It violates best practices for scalability and reliability. Instead, use a managed service like Vertex AI or Cloud Run for serving the model.