Domain 4: Serve and scale models

Google Cloud ML Engineer · this domain is approximately 17.4% of the exam · 0 practice questions.

The Serve and scale models domain focuses on deploying trained models to production using Vertex AI Prediction, including online prediction endpoints, batch prediction jobs, and the configuration of autoscaling to handle variable traffic loads. For the Google Cloud ML Engineer exam, candidates must understand how to optimize serving infrastructure for latency and cost, including model export formats, container images for custom prediction, and traffic splitting for A/B testing or canary deployments. This domain also covers when to use batch versus online prediction and how to structure model endpoints to meet service level objectives.

Practice all 0 questions in this domain

The full Google Cloud ML Engineer bank includes 0 more questions in this domain, each with a verified answer and a written explanation.