A company uses Databricks Asset Bundles to deploy ML assets across dev, staging, and production environments. They need to roll back a model serving endpoint to a previous model version after a bad deployment. What is the most efficient way to achieve this?

Delete the current bundle and recreate it with the previous configuration.

Manually update the serving endpoint in the UI to point to the previous model version.

Use the MLflow API to transition the previous model version back to 'Production'.

Redeploy the previous bundle version using `databricks bundle deploy --environment prod`.

Which best practice is most important when designing a Databricks environment for a team of data scientists working on multiple ML projects?

Assign all users the same permissions to simplify collaboration.

Organize workspaces with folders and access controls to separate projects and teams.

Create separate workspaces for each project to isolate resources and permissions.

Use a single shared workspace for all projects to simplify administration.

A company uses Unity Catalog to manage data access. Data engineers and data scientists need different permissions on the same Delta tables. The data scientists should only read a subset of columns, while data engineers need full access. What is the recommended way to implement this?

Use a cluster policy to restrict which columns data scientists can see in the table.

Grant the data scientists SELECT on the table and rely on them to only query the allowed columns.

Create two separate Delta tables, one for each team, and sync data periodically.

Create a view that exposes only the allowed columns and grant the data scientists SELECT on the view, while granting data engineers SELECT on the underlying table.

Which of the following ML assets can be defined in a Databricks Asset Bundle?

Only registered models and model serving endpoints.

Only MLflow experiments and registered models.

MLflow experiments, registered models, and model serving endpoints.

Only MLflow experiments and model serving endpoints.

DatabricksCertified Machine Learning Professional

Domain 2Objective 3

Environment Architectures MACHINE-LEARNING-PROFESSIONAL Practice Questions (Page 3)

Part of the ML Ops domain, which accounts for 44% of the MACHINE-LEARNING-PROFESSIONAL exam. Databricks does not publish an official question count, but from its 120-minute exam (~50–80 total, ~22–35 in this domain), expect 4–7 from this objective — we provide 28 practice questions to prepare you well beyond it. (estimate)

28questions here

6free pages

9concepts

44%of the exam

Questions 11–15

11expert · hard
A company uses Databricks Asset Bundles to deploy ML assets across dev, staging, and production environments. They need to roll back a model serving endpoint to a previous model version after a bad deployment. What is the most efficient way to achieve this?
Select an answer first
12foundation · easy
Which best practice is most important when designing a Databricks environment for a team of data scientists working on multiple ML projects?
Select an answer first
13application · medium
A company uses Unity Catalog to manage data access. Data engineers and data scientists need different permissions on the same Delta tables. The data scientists should only read a subset of columns, while data engineers need full access. What is the recommended way to implement this?
Select an answer first
14foundation · easy
In a Databricks Asset Bundle, which field in the 'mlflow_experiment' resource specifies where artifacts are stored?
Select an answer first
15foundation · easy
Which of the following ML assets can be defined in a Databricks Asset Bundle?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to MACHINE-LEARNING-PROFESSIONAL

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. “MACHINE-LEARNING-PROFESSIONAL” is a trademark of its owner, used for identification only.