A company runs a document-summarization workload on Databricks. Users submit documents that are often similar or identical to previously submitted ones. The team wants to reduce token consumption and cost without changing the quality of the summaries. Which strategy should they implement?

Increase the max tokens parameter for the model to allow longer outputs.

Implement a caching layer that stores and reuses responses for identical or similar prompts.

Reduce the number of serving endpoints to one shared endpoint.

Switch to a larger model to improve accuracy.

A team is using Databricks Model Serving and wants to set up a budget alert to notify them when monthly spending exceeds a threshold. Which approach is the most reliable?

Manually check the billing page in the cloud provider console.

Create a Databricks SQL dashboard that queries the system tables for usage and set up an alert on the total cost.

Set up a cluster policy to limit spending.

Use the MLflow UI to monitor costs.

A company runs a customer-support chatbot on a Databricks Model Serving endpoint using a large LLM. The endpoint is configured for a provisioned throughput of 10 concurrent requests. After a traffic analysis, the team finds that the average concurrency is only 2, and the endpoint is idle for most of the night. They want to reduce costs without sacrificing availability during peak hours. What should they do?

Switch the endpoint to a serverless model serving configuration and rely on its automatic scaling.

Move the model to a batch inference job that runs nightly.

Keep the provisioned throughput but reduce the concurrency to 2 to match the average.

Enable caching on the endpoint to serve repeated queries without invoking the model.

A company runs a multi-tenant application where different customers have different usage patterns. They want to allocate costs accurately to each customer. They have enabled system tables for usage. What is the best way to track cost per customer?

Estimate costs based on the number of users per customer.

Include a customer identifier in the request metadata and use the usage system table to aggregate token usage by that identifier.

Create a separate serving endpoint for each customer.

Use the cluster logs to identify customer requests.

DatabricksCertified Generative AI Engineer Associate

Domain 6Objective 4

Cost Control and Usage Tracking GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 2)

Part of the Section 6: Evaluation and Monitoring domain, which makes up ~17% of our current practice bank. Databricks does not publish an official question count, but from its 90-minute exam (~35–60 total, ~6–10 in this domain), expect 1–2 from this objective — we provide 21 practice questions to prepare you well beyond it. (estimate)

21questions here

5free pages

3concepts

Questions 6–10

6foundation · easy
Which Databricks model serving option is most appropriate for a workload with highly variable traffic patterns, where the team wants to avoid paying for idle capacity?
Select an answer first
7application · medium
A company runs a document-summarization workload on Databricks. Users submit documents that are often similar or identical to previously submitted ones. The team wants to reduce token consumption and cost without changing the quality of the summaries. Which strategy should they implement?
Select an answer first
8expert · medium
A team is using Databricks Model Serving and wants to set up a budget alert to notify them when monthly spending exceeds a threshold. Which approach is the most reliable?
Select an answer first
9application · medium
A company runs a customer-support chatbot on a Databricks Model Serving endpoint using a large LLM. The endpoint is configured for a provisioned throughput of 10 concurrent requests. After a traffic analysis, the team finds that the average concurrency is only 2, and the endpoint is idle for most of the night. They want to reduce costs without sacrificing availability during peak hours. What should they do?
Select an answer first
10expert · medium
A company runs a multi-tenant application where different customers have different usage patterns. They want to allocate costs accurately to each customer. They have enabled system tables for usage. What is the best way to track cost per customer?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to GENERATIVE-AI-ENGINEER-ASSOCIATE

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 Databricks. “GENERATIVE-AI-ENGINEER-ASSOCIATE” is a trademark of its owner, used for identification only.

Cost Control and Usage Tracking GENERATIVE-AI-ENGINEER-ASSOCIATE Practice Questions (Page 2)

Questions 6–10

Which Databricks model serving option is most appropriate for a workload with highly variable traffic patterns, where the team wants to avoid paying for idle capacity?

A company runs a document-summarization workload on Databricks. Users submit documents that are often similar or identical to previously submitted ones. The team wants to reduce token consumption and cost without changing the quality of the summaries. Which strategy should they implement?

A team is using Databricks Model Serving and wants to set up a budget alert to notify them when monthly spending exceeds a threshold. Which approach is the most reliable?

A company runs a multi-tenant application where different customers have different usage patterns. They want to allocate costs accurately to each customer. They have enabled system tables for usage. What is the best way to track cost per customer?

Keep studying — Section 6: Evaluation and Monitoring