Databricks certification path

Start with the associate exam for the job you actually do — Data Analyst, Data Engineer, Machine Learning or Spark Developer — and only then consider a professional exam. The professional exams are a genuine step up, not a slightly harder version of the associate.

The seven exams

Pick the lane, not the ladder

These are not seven rungs. They are three or four separate lanes, each with an associate entry point and, in two cases, a professional exam above it.

• Analytics lane — Data Analyst Associate. No professional equivalent.
• Engineering lane — Data Engineer Associate, then Professional.
• Machine learning lane — ML Associate, then Professional.
• Generative AI — GenAI Engineer Associate, which sits alongside the ML lane rather than above it.
• Spark itself — Spark Developer Associate, which is about the API rather than the platform.

Collecting all seven is not a strategy. Two well-chosen certifications in your lane say more than seven scattered across lanes you do not work in.

Where Spark Developer fits

The Spark Developer Associate is the odd one out. It tests the DataFrame API, Spark architecture, Spark SQL and tuning — knowledge that underpins the data engineering exams but is not the same thing.

It is worth taking if you write Spark code directly and want to prove that specific skill, or if you found the Data Engineer Associate hard because the Spark fundamentals were shaky. It is not a prerequisite, and it is not a lighter version of the engineering exam.

How big is the jump to professional?

Bigger than most people expect. The associate exams ask which feature or operation fits a situation. The professional exams assume you know that and ask what you would do when it goes wrong in production.

The Data Engineer Professional, for example, puts over a fifth of its weight on developing code for data processing, and covers cost and performance optimisation, monitoring and alerting, debugging and deployment, security and governance. The questions describe situations you only recognise if you have been on call for a pipeline.

The ML Professional splits its weight almost entirely between model development and MLOps, with deployment a smaller slice. Again, the MLOps half rewards people who have actually operated models rather than trained them.

Rule of thumb: give yourself six to twelve months of production work on the platform between the associate and the professional exam in the same lane.

One thing nobody publishes

Databricks does not publish passing scores for these exams. If you find a site quoting a specific percentage, it is guessing or repeating someone else’s guess. Treat any confident number with suspicion — including in study material you have paid for.

That makes practice tests more useful than usual here, because a score is the only signal you get before the real thing.

How long each takes

For someone already working on the platform:

• Associate exams — four to six weeks
• Professional exams — eight to twelve weeks, and that assumes the production experience is already there

Without daily platform work, these estimates roughly double, and the professional exams become impractical rather than merely hard.

What to do next

Choose your lane, read the exam format guide for the associate exam in it, and take the free 20-question practice test in that section. It is the fastest way to find out whether you are closer to ready than you think.

Weighing Databricks against a Microsoft route? See Databricks vs Azure for data engineering. The exam changes tracker records which versions are current.