Databricks ML Professional
Databricks Certified Machine Learning Professional
More study resources
Cheat sheet
Keep the key ideas within reach.
ExploreKey terms
Get comfortable with the vocabulary.
ExploreStudy plan
Give each week a clear focus.
ExploreCommon mistakes
Understand the traps and how to avoid them.
ExploreCase studies
Put your knowledge to work in context.
ExploreExam-day guide
Prepare for the day ahead.
ExploreAbout this exam
The Databricks Certified Machine Learning Professional is a proctored certification exam that assesses a candidate's ability to design, implement, and manage enterprise-scale machine learning solutions using advanced Databricks platform capabilities. It sits above the associate-level ML credential and targets practitioners who already build production ML systems: data scientists and ML engineers working with SparkML pipelines, distributed training and hyperparameter tuning, advanced MLflow features, and Feature Store concepts for automated feature pipelines. The exam also probes MLOps practice - testing strategies, environment management with Declarative Automation Bundles, automated retraining workflows, and drift monitoring with Lakehouse Monitoring - plus deployment topics such as custom model serving and model rollout management.
Scoring weight concentrates in two domains: Model Development (44%) and ML Ops (44%), with Model Deployment at 12%. The exam is delivered in English, online or at a test center, with no test aides allowed, and it assesses SQL ability against ANSI SQL standards. There are no prerequisites, though related training is strongly recommended, and Databricks suggests roughly one or more years of hands-on experience with the tasks in the exam guide. Certification is valid for two years, after which recertification on the current exam version is required.
Preparation should start from the official Exam Guide, which lists the tasks and related training. Candidates should then practise the workflows hands-on - building and tuning pipelines, tracking runs in MLflow, defining feature tables, writing bundle-based deployment configs, and wiring monitoring and retraining jobs - before reviewing the guide again to close gaps. This page is independent study material and is not affiliated with or endorsed by Databricks.
What you'll study
About Databricks ML Professional preparation
How many questions are on the exam and how long do I get?
There are 59 scored questions and the time limit is 120 minutes, which works out to roughly two minutes per question. The exam may also contain unscored items used for statistical purposes; these are not identified and do not affect your score, and additional time is factored in to account for them.
What does the exam cost and where do I take it?
Registration is $200. You can take the exam online or at a test center, and it is proctored either way. It is delivered in English. Registration happens through the exam delivery platform linked from the official certification page, where you log in or create an account.
Which domain should I study first?
Model Development and ML Ops each carry 44%, so together they are 88% of the exam, and Model Deployment is 12%. Start with the two heavy domains and treat deployment as a final pass. Within Model Development, prioritise SparkML pipelines, distributed training and tuning, MLflow, and Feature Store, since those are the named capabilities.
How difficult is it, and what experience do I need?
There are no prerequisites, but Databricks recommends 1+ years of hands-on experience performing the machine learning tasks outlined in the exam guide. This is a professional-level exam, so it assumes working fluency rather than introductory knowledge. Candidates without production Databricks ML experience should expect to build that familiarity through practice before sitting it.
What are the most common failure modes?
The two largest domains reward depth over recall: candidates who have read about MLflow tracking or Feature Store but never configured them tend to struggle. Weak SQL is another risk, since the exam assesses SQL and adheres to ANSI SQL standards. Skipping the Exam Guide and studying generic ML theory instead of Databricks-specific workflows is a frequent mistake.
How long is the certification valid, and how do I renew?
The validity period is two years. Recertification is required every two years to maintain certified status, and to recertify you must take the current version of the exam. Databricks advises reviewing the Getting Ready for the Exam section to prepare for the recertification attempt rather than assuming the older exam content still applies.
Can I use notes, documentation, or a second monitor?
No. Test aides are not allowed. The exam is proctored, so prepare to work from memory on APIs, configuration patterns, and SQL. Because the exam assesses SQL against ANSI SQL standards, practise writing queries without an assistant or autocomplete so the syntax is automatic under time pressure.
How should I structure my final week of study?
Reread the Exam Guide and list every task you cannot perform unaided, then close those gaps with hands-on work rather than more reading. Run a system check and review the technical requirements before exam day. Databricks also publishes an AI Prep Guide, and the related training listed in the Exam Guide is the recommended baseline.
Is this page an official Databricks resource?
No. This is independent study material written for candidates searching for this exam. It is not affiliated with, endorsed by, or produced by Databricks. Always confirm exam details, registration steps, and the current Exam Guide on the official Databricks certification page, since exam versions and logistics can change.