A real-time translation application needs to respond to user requests in under 500 milliseconds. The application currently uses Gemini 1.5 Pro with a long system prompt. Which change would most effectively reduce latency?

Add more detailed instructions to the system prompt to improve accuracy

Increase the number of output tokens to allow more complete translations

Use a batch API to process multiple requests at once

Switch to Gemini 1.5 Flash and shorten the system prompt

A team is choosing between two Gemini models in Model Garden for a real-time chat application. The application has strict latency requirements and users typically send short messages. Which model selection criterion is most relevant?

The model's training data cutoff date

The model's maximum output token limit

The model's inference speed and size

The model's ability to generate images

A developer notices that a Gemini-based application is sending the same large system prompt with every request. Which strategy would most effectively reduce token usage and cost?

Disable caching to ensure fresh responses

Increase the max output tokens to allow longer responses

Switch to a larger model tier for better quality

Use prompt caching to store the system prompt and reuse it across requests

A company is building a customer support chatbot using Gemini. The chatbot handles a high volume of similar questions. They want to reduce costs without significantly affecting response quality. Which strategy is most effective?

Implement a caching layer for common queries and use Gemini 1.5 Flash for new ones

Use Gemini 1.5 Pro for all queries to ensure the highest quality responses

Send the entire conversation history with every request to provide context

Increase the max output tokens to allow more detailed responses

A developer wants to reduce the latency of a Gemini-based application that generates long responses. Which prompt engineering technique is most likely to help?

Instruct the model to provide concise answers

Increase the max output tokens to allow longer responses

Disable caching to ensure fresh responses

Use a more complex prompt with multiple examples

Google CloudProfessional Machine Learning Engineer

Domain 1Objective 2

Building AI Solutions Using Google Cloud AI APIs or Foundational Models PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 3)

Part of the Architecting low-code AI solutions domain, which accounts for ~13% of the PROFESSIONAL-MACHINE-LEARNING-ENGINEER exam.

16questions here

4free pages

6concepts

~13%of the exam

Questions 11–15

11application · medium
A real-time translation application needs to respond to user requests in under 500 milliseconds. The application currently uses Gemini 1.5 Pro with a long system prompt. Which change would most effectively reduce latency?
Select an answer first
12foundation · easy
A team is choosing between two Gemini models in Model Garden for a real-time chat application. The application has strict latency requirements and users typically send short messages. Which model selection criterion is most relevant?
Select an answer first
13foundation · easy
A developer notices that a Gemini-based application is sending the same large system prompt with every request. Which strategy would most effectively reduce token usage and cost?
Select an answer first
14application · medium
A company is building a customer support chatbot using Gemini. The chatbot handles a high volume of similar questions. They want to reduce costs without significantly affecting response quality. Which strategy is most effective?
Select an answer first
15foundation · easy
A developer wants to reduce the latency of a Gemini-based application that generates long responses. Which prompt engineering technique is most likely to help?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to PROFESSIONAL-MACHINE-LEARNING-ENGINEER

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Google Cloud. “PROFESSIONAL-MACHINE-LEARNING-ENGINEER” is a trademark of its owner, used for identification only.