Which Databricks certifications cover data engineering and machine learning?
Databricks groups its certifications by role. The data engineer track has two levels: Data Engineer Associate for introductory pipeline work and Data Engineer Professional for advanced production work. The ML engineer track works the same way, with Machine Learning Associate for basic ML tasks and Machine Learning Professional for production ML at enterprise scale. All four are proctored, multiple-choice exams that you can take online or at a test center. None of them has a formal prerequisite, though Databricks strongly recommends its related training.
The catalog also lists related exams that touch this work: Generative AI Engineer Associate, a Context Engineering Associate for AI agent systems, an Associate Developer for Apache Spark covering the DataFrame API, and a renewal exam for Data Engineer Associate. Databricks retired its Hadoop Migration Architect exam on August 1, 2024. If you see it on an old study list, skip it.
Data Engineer Associate vs. Professional
The associate exam has 45 scored questions in 90 minutes. It tests ingestion and loading (21%), transformation and modeling (22%), Lakeflow Jobs (16%), governance and security (15%), CI/CD (10%), troubleshooting and optimization (10%) and platform basics (6%). It expects ETL with PySpark. Data code appears in SQL where possible and in Python otherwise. It suits someone with hands-on experience doing the pipeline tasks listed in Databricks' exam guide.
The professional exam has 59 scored questions in 120 minutes. Python and SQL coding is the biggest domain at 22%, followed by cost and performance optimization at 13%. It covers Delta Lake, Unity Catalog, Auto Loader, streaming, Medallion Architecture, and deployment with the Databricks CLI, REST API and Asset Bundles. Pick it once you own production pipelines and care about cost, monitoring and security.
Machine Learning Associate vs. Professional
The ML associate exam has 48 scored questions in 90 minutes. It weights Databricks Machine Learning at 38%, model development at 31%, ML workflows at 19% and deployment at 12%, with tools such as AutoML, Unity Catalog and parts of MLflow. Databricks recommends six or more months of hands-on experience. All ML code is in Python.
The ML professional exam has 59 scored questions in 120 minutes and is offered in English only. Model development and MLOps each carry 44%, and deployment carries 12%. It covers SparkML pipelines, distributed training and tuning, Feature Store, drift monitoring with Lakehouse Monitoring, and model rollout. Databricks recommends at least a year of hands-on experience. It fits engineers who deploy, monitor and retrain models in production.
Who should bother, and what a certification does not do
These certifications prove you know one vendor's platform. They help most if you work for a Databricks customer or partner, consult on Databricks projects, or want to show platform skills when switching from another stack. They will not get you hired by themselves. Interviewers still test SQL, Python, data modeling and system design. If your target employers do not run Databricks, put the time into general skills and portfolio work first. Start at associate level unless you already ship production workloads on the platform.
Pay
USD a year, before tax. US figures. BLS OEWS May 2025, moved to 2026 with the Employment Cost Index, using database architects, the nearest series to data engineers; it covers the whole occupation, including people outside AI and Databricks work. Job ad rows are midpoints of pay ranges in 31 live AutomationRoles.ai ads from 26 employers on 5 November 2026; they lean toward large AI and tech employers and show base pay only, leaving out stock and bonus. Refer to the Data & AI Engineers salary guide for figures from live adverts.
What you need
- PrerequisitesNone of the four exams has a formal prerequisite. Databricks highly recommends its related training but does not require it.
- AgeDatabricks says its certifications are for people aged 18 and over.
- ExperienceDatabricks recommends 6+ months of hands-on work for ML Associate and 1+ year for ML Professional. The data engineer exams ask for hands-on experience with the tasks in the exam guide.
- FeeEach exam costs $200. Every retake is charged at the full price, and Databricks does not give free retake vouchers.
- FormatProctored, multiple choice, no test aids. Taken online or at a test center, with registration through the Webassessor platform. Exams may include unscored questions, with extra time allowed for them.
- RenewalCertifications are valid for two years from the award date. To recertify, you pass the current version of the exam.
How to get there
- Months 0-6Learn the platform basicsGet hands-on time in Databricks with SQL, Python and PySpark on real or practice data.
- Month 6Choose a trackPick the data engineer or ML engineer track based on the work you do or want to do.
- Months 6-9Pass the associate examStudy from the current exam guide, which Databricks calls the source of truth, then sit the 90-minute exam.
- Months 9-18Build production experienceOwn pipelines or models in production, including monitoring, cost tuning and CI/CD.
- Year 1-2Pass the professional examSit the 120-minute, 59-question exam once you have about a year of production work.
- Every 2 yearsRecertifyPass the current version of the exam before the two-year validity runs out.
Open now on AutomationRoles.ai
Questions people ask
How much does a Databricks certification exam cost?
Every Databricks certification exam costs $200. Retakes cost the same as the first attempt, and Databricks does not offer free retake vouchers.
How long is a Databricks certification valid?
Two years from the date it is awarded. To stay certified, you pass the current version of the exam, and Databricks notifies you before your certification expires.
Should I start with associate or professional?
Start with associate unless you already run production pipelines or models on Databricks. Databricks recommends at least a year of hands-on experience for the ML Professional exam, and the professional exams are longer, with 59 questions in 120 minutes.
What happens if I fail?
You can register again and retake the exam after 14 days. There is no limit on attempts, but you wait 14 days between each one and pay full price each time.
Does Databricks offer official practice exams?
Databricks' FAQ says it does not offer practice exams. It points candidates to the exam guide on each certification page, which lists the current topics and objectives.
Is there a Databricks certification for generative AI work?
Yes. The Generative AI Engineer Associate has 45 questions in 90 minutes, costs $200 and covers building RAG applications and LLM chains on Databricks. Databricks recommends six or more months of hands-on experience before taking it.
Sources
- Databricks: Data Engineer Associate certification
- Databricks: Data Engineer Professional certification
- Databricks: Machine Learning Associate certification
- Databricks: Machine Learning Professional certification
- Databricks: Generative AI Engineer Associate certification
- Databricks: Certification and badging catalog
- Databricks: Certification and badging FAQ
