Model Deployment Questions
Practice questions for Model Deployment topic in Databricks Certified Machine Learning Professional. 24 questions covering this domain.
An ML engineer updates a custom model serving endpoint to serve a new model version. How does Databricks handle the transition to minimize risk to liv...
An ML team wants to deploy a PyTorch transformer model for real-time inference. The model weighs 8 GB and requires GPU acceleration. During peak hours...
Which model types can be deployed using Mosaic AI Model Serving?
A data scientist deploys a custom pyfunc model to Mosaic AI Model Serving. During deployment, the endpoint shows a dependency error for a private inte...
An ML team deploys a real-time model serving endpoint and enables the scale-to-zero feature. A product manager is concerned about latency for the firs...
What is Mosaic AI Model Serving on Databricks?
Which Mosaic AI Model Serving compute type is appropriate for deploying a CPU-based scikit-learn model that expects low and irregular request volume?
What is the purpose of the `/invocations` endpoint on a Mosaic AI Model Serving endpoint?
An ML team configures A/B testing on a Mosaic AI Model Serving endpoint by splitting traffic 80% to the `champion` model (version 5) and 20% to the `c...
A data scientist configures a Mosaic AI Model Serving endpoint and enables the inference table feature. The endpoint receives requests but the inferen...
A data scientist deploys a model with Mosaic AI Model Serving and enables automatic feature lookup from the Databricks Online Feature Store backed by ...
An ML team deploys a challenger model alongside the champion model on the same Mosaic AI Model Serving endpoint using traffic splitting (90% champion,...
A platform team wants the lower-latency benefits of route optimization on a new custom model endpoint. When can they enable route optimization?
An endpoint is expected to handle roughly 12 queries per second, and each model run takes about 0.5 seconds. Which documented rule of thumb should gui...
A custom model endpoint accepts tabular input from a pandas DataFrame. Which request format does Databricks recommend, and why?
A team wants one endpoint to serve a custom MLflow model and an external OpenAI model behind the same traffic split. What does Databricks support?
Which authentication method is supported for querying route-optimized model serving endpoints?
A multi-model endpoint routes 90 percent of traffic to current and 10 percent to challenger. A tester wants every request to hit challenger directly a...
A model serving endpoint was created by a user identity that lacks permission to read the Unity Catalog model the endpoint must deploy. The team wants...
An engineer updates a custom-model endpoint to a new configuration and wants to avoid dropping live traffic during the rollout. How does Databricks ha...
Sign in to see all 24 questions
Create a free account to browse all questions — completely free during our launch phase.