Which Databricks certification? All 8 compared for 2026

There are 8 Databricks certification exams in 2026, and the right first one is the associate exam for the work you already do: Data Engineer Associate for pipelines, Data Analyst Associate for SQL and dashboards, Machine Learning Associate for models, Generative AI Engineer Associate for RAG apps, and the new Context Engineer Associate for agents. Every exam costs $200 and lasts 2 years. The two professional exams, Data Engineer and Machine Learning, are worth it only after about a year of the same work.

The list changed a lot this year. Context Engineer Associate is new, its guide published on 29 July 2026. The Apache Spark Developer and Data Engineer Professional exams got new guides in October, and Data Engineer Associate gained sections on CI/CD and on troubleshooting in August. Advice written before 2026 describes exams that no longer exist in that form.

The 8 exams side by side

All 8 are multiple choice, with no notes or documentation allowed, taken online with a webcam proctor or at a test centre. The question counts are scored questions. On the Context Engineer and Spark exams, up to 10 more questions may appear that Databricks is trying out; they do not count, and you cannot tell which ones they are, so answer all of them.

Pick the first one by the work you do

Try itWhich Databricks certification first

Your work

·Build data pipelines·Write SQL, dashboards and Genie spaces·Train and deploy ML models·Build RAG and GenAI apps·Build AI agents·Write Spark code, on any platform

New to it6 months or soA year or more

1. Step 1, $200

   Data Engineer Associate

   Exam

   45 scored questions, 90 min

   Weighs most

   Data Transformation and Modeling, 22%

   Languages

   English, Japanese, Portuguese, Korean

Worth adding later: Apache Spark Developer Associate, the exam next to this work.

The widget's paths are our advice, not a Databricks rule. Databricks does not require any exam before another, so you can book Data Engineer Professional first if you want to. What it does recommend is hands-on time: about 6 months for most associate exams, and a year or more for Machine Learning Professional.

What each exam is really testing

Data Engineer Associate is the broadest exam and the most common first step. The August 2026 guide spreads it over 7 sections: ingestion (21%) and transformation (22%) carry the most, Lakeflow Jobs 16%, and governance with Unity Catalog 15%. The 2 newer sections, CI/CD and troubleshooting at 10% each, are where older study material falls short. Our Data Engineer Associate guide goes through it section by section.

Data Engineer Professional is mostly about code. In the October 2026 guide the largest section, 23%, is writing Python and SQL for data processing, and cost and performance tuning follows at 15%. Ingestion, which now includes sharing and federation, and data manipulation are 12% each, monitoring and debugging 10% each, and security, governance and modelling the rest. It suits someone who already runs production pipelines and wants proof of it. Our Data Engineer Professional study guide shows what moved from the earlier outline, which the certification page still shows.

Data Analyst Associate is the SQL exam. A fifth of it is running queries on Databricks SQL and choosing warehouses, then dashboards (16%), reading query performance (15%) and Genie spaces (12%), where business users ask questions in plain language. It is offered in English only.

Machine Learning Associate and Professional split cleanly. The associate exam is about the platform: 38% on Databricks Machine Learning itself, the rest on workflows, model development and deployment. The professional exam is 44% model development and 44% ML Ops, the work of keeping models trained, tracked and served.

Generative AI Engineer Associate is about building an application: designing it (14%), preparing and chunking the data (14%), writing the application (30%), assembling and deploying it (22%), then governance and evaluation. It is the exam for people who build retrieval apps on Databricks, and our Generative AI Engineer Associate study guide turns its 6 sections into one project.

Context Engineer Associate is the 2026 addition, and its sections read like a map of how agents fail. Retrieval and Genie configuration are 20%, memory and state in Lakebase and MLflow 18%, tools and MCP 13%, compaction 11% and multi-agent, long-running work 13%. If you build agents, this tests the work more closely than the Generative AI exam does. Our Context Engineer Associate study guide shows how its questions are built, and Agent MemoryAgent MemoryThe ways an agent keeps what it needs beyond a single context window, such as summaries, notes in files, and saved facts it reads back later. The model itself remembers nothing. covers one of its biggest sections.

Apache Spark Developer Associate is the one exam that is mostly about open-source Spark rather than the Databricks platform. The DataFrame API is 30%, Spark SQL and architecture 20% each, and Spark Connect and the pandas API on Spark are 5% each in the October 2026 guide. Code is in Python. It suits Spark developers on any platform, which also makes it the easiest certificate to explain outside Databricks shops. Our Spark Developer Associate study guide lists the method pairs it tests, and the Spark UI guide covers much of its tuning section.

What a certification costs over the years

The exam fee is $200 for each exam, plus local tax where it applies. A fail costs another $200, and you wait 14 days before the next attempt.

The cost people forget is renewal. A certification lasts 2 years, and you renew it by passing the current version of the exam again, at full price. Data Engineer Associate is the one exam with a shorter renewal exam: 25 questions in 60 minutes, online only, still $200. So one certification kept for 4 years costs $400 if you pass every time, and 2 kept for 4 years cost $800.

Two things follow. Renewing every exam you ever passed gets expensive, so keep the ones your work and your CV use. And because you renew by sitting the current exam, a certificate earned in 2024 says nothing about the 2026 sections until you renew it.

The passing score, and why it is not published

Databricks does not publish a passing score. Its FAQ says the score is set through statistical analysis and can change. Any number you see quoted online is a guess from people who passed, so treat it as unknown and aim to know every task in the guide.

The free badges are the exception. Fundamentals assessments, and the platform architect and administrator badges, pass at 80% and cost nothing.

Before you pay $200

Start with the free material Databricks gives away:

• The exam guide. Each exam's page links a PDF guide with every section, its weight and the tasks it tests. It is the only outline you can rely on, and our table above is taken from those guides. Check its date before you study; the Spark and Data Engineer Professional guides changed this October.
• A free badge. Databricks Fundamentals, Generative AI Fundamentals, Analytics Fundamentals and AI Agent Fundamentals each have on-demand training and a free assessment. They are a way to check whether the platform's vocabulary is familiar before you book a paid exam.
• Databricks Free Edition. The practical tasks behind most sections can be tried in Free Edition without a paid workspace. Our data engineering course runs on it.

Then book the exam about 2 weeks after you can explain every task in its guide without notes. The 14-day retake wait means a failed attempt costs a fortnight as well as $200.

Our picks, in one line each

• You build pipelines: Data Engineer Associate, then Professional after a year. Add Spark Developer if you work on Spark outside Databricks too.
• You write SQL and dashboards: Data Analyst Associate. It is the only one aimed at you.
• You train models: Machine Learning Associate, then Professional.
• You build RAG or GenAI apps: Generative AI Engineer Associate, then Context Engineer.
• You build agents: Context Engineer Associate first. It tests the agent work most directly.
• You are new to Databricks: a free Fundamentals badge, then the associate exam for your role.

If you build on Databricks and want the context behind these exams rather than only the questions, our AI engineering course and the data engineering subject are free and follow the same order.