Chapter 4 of 4 · Databricks Machine Learning Associate
Model Deployment
Batch, streaming and real time serving, pandas inference, custom models and traffic splits.
- 2topics
- 10 minfast track
- 12%of the exam
Topics in this chapter
Take them in order, or jump to the one you need.
- 1Batch, Streaming and Real Time InferenceDatabricks ML Associate: choose batch, streaming or real time inference, score a pandas DataFrame with MLflow, and apply a model…5 min
- 2Model Serving Endpoints and Traffic SplittingDatabricks ML Associate: deploy a custom pyfunc model to a Model Serving endpoint, query it over REST, and split traffic between…5 min
About this chapter
What this chapter covers
Model Deployment is 12% of the Databricks Machine Learning Associate exam, the smallest chapter. It covers how a registered model is put to work: scoring data in batches, scoring a stream, and answering requests in real time.
The chapter has two parts: choosing between batch, streaming and real time serving (including batch inference with pandas and streaming inference in a Delta Live Tables pipeline), and deploying, querying and splitting traffic on Mosaic AI Model Serving endpoints, including custom pyfunc models.
Questions are single-answer and multiple-selection ("Which TWO ...?") questions, mostly short scenarios. Some show a short Python snippet (MLflow, Feature Engineering client, Spark ML, Model Serving) and ask what it does or which call meets a requirement. There are no exhibits.
How the topics fit together
Every deployment starts from a model registered in Unity Catalog. The latency requirement then decides whether you score a table on a schedule, score events as they arrive, or put the model behind a REST endpoint.
- Approach questions ask which of batch, streaming or real time fits a stated latency, volume and cost.
- Batch questions ask how to score a pandas DataFrame or a Spark DataFrame with an MLflow model.
- Streaming questions ask how a model is applied inside a Delta Live Tables pipeline.
- Endpoint questions ask how to deploy a custom model, how to query it and what the request body looks like.
- Traffic questions ask how to split requests between two models on one endpoint.
Look for the latency words first. "Nightly" or "once a day" means batch, "events arriving continuously" means streaming, and "per request in milliseconds" means a Model Serving endpoint.
How to use these pages
Each topic page has the same shape, so you can skim for what you need.
- What it is: the idea in plain words.
- How it works: the mechanism, with tables and examples.
- What to remember for the exam: the facts to memorize and how questions are phrased.
- Common mistakes: the classic traps.
- Check yourself: three questions to test your understanding.
Ready for more practice?
The full Databricks Machine Learning Associate question bank has 600 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Databricks. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.