Model Deployment Questions
Practice questions for Model Deployment topic in Databricks Certified Machine Learning Associate. 24 questions covering this domain.
In the Unity Catalog Model Registry, what replaces the stage concept (Staging, Production, Archived) that exists in the Workspace Model Registry?
To configure MLflow to use Unity Catalog as the model registry backend instead of the Workspace Model Registry, which configuration call is required?
A data scientist has registered a model in the Workspace Model Registry and wants to expose it as a REST API for real-time inference. Which Databricks...
A data science team deploys a model for batch inference using `mlflow.pyfunc.spark_udf()` on a DataFrame with 1 million rows. The job is very slow des...
A data science team wants to automatically trigger a CI/CD job when a model version is transitioned to Production in the Workspace Model Registry. Whi...
A junior data scientist requests a stage transition to Production in the Workspace Model Registry but does not have the required permissions. What hap...
A data scientist registers a model in the Unity Catalog Model Registry and assigns the alias `champion` to version 5. Later, version 6 is trained and ...
Which URI scheme is used to reference a model version by alias in the Unity Catalog Model Registry?
What does a Mosaic AI Model Serving endpoint return when called with a POST request?
A data scientist creates a new challenger model and wants to test it alongside the existing production model in Mosaic AI Model Serving without changi...
A data science team uses Mosaic AI Model Serving to deploy a custom PythonModel. During load testing, the endpoint returns HTTP 429 errors during peak...
A data scientist wants to register a model in the Unity Catalog Model Registry using the MLflow Python API after training. Which combination of config...
Which models does Databricks classify as custom models for Model Serving?
A team wants one serving endpoint to expose both a current production model and a challenger version while controlling request percentages. Which endp...
An endpoint creator no longer has the Unity Catalog permissions needed by a served model. What must the team do to change the identity used by that en...
Why is scale to zero a poor choice for latency-sensitive production endpoints?
A production service calls a Databricks custom model endpoint through the REST API. What authentication practice does Databricks recommend?
When sending a pandas-style JSON request to a custom model serving endpoint, which format does Databricks recommend?
Which models count as custom models for Databricks Model Serving?
Why is scale to zero a poor choice for a latency-sensitive serving endpoint?
Sign in to see all 24 questions
Create a free account to browse all questions — completely free during our launch phase.