Databricks Data Engineer: Professional vs Associate

The Associate covers introductory data engineering on Databricks. The Professional covers advanced work. The Associate is not a prerequisite — you can sit the Professional first — but the progression is the obvious one.

Side by side

Same price, same validity, same format. The Professional is longer and pitched harder. Professional figures are for the outline that applies from 9 October 2026; see the exam format for what changed.

What grows, what shrinks

Code becomes the largest section. Developing Code for Data Processing using Python and SQL is 23% of the Professional and has no Associate equivalent. The Associate tests platform operation; the Professional tests whether you can read and reason about the code doing the work.

Optimisation gets its own section. Cost & Performance Optimisation at 15%. The Associate folds this into a combined 10% troubleshooting section. At Professional level, tuning is a discipline in its own right.

Ingestion shrinks — from 21% to 12% — and changes character. The Professional assumes you can load files and asks about CDC from relational databases, OpenSharing (formerly Delta Sharing), Clean Rooms and Lakehouse Federation instead.

New sections appear: Data Manipulation (12%), with VARIANT and AI functions, and a distinct Data Modelling section (5%), neither of which the Associate has in that form.

Everything becomes more granular. Nine sections rather than seven, three of them 8% or less. Blind spots are harder to hide because the exam samples more places.

The real difference in one sentence

The Associate asks can you build a pipeline? The Professional asks can you write it well, make it cheap and fast, observe it, secure it, and fix it when it breaks?

Which should you take?

Associate if you are newer to Databricks, or you use the platform without owning production pipelines. It is the natural foundation and its content is assumed by the Professional.

Professional if you write and tune production Spark and SQL, care about cluster cost, and are the person paged when a pipeline fails. The Associate will feel like a formality.

Both, in order, if you are building a career here. The Associate is cheaper in effort and genuinely preparatory.

Can you go straight to Professional?

Yes, and for experienced engineers it is often the right call. The test is a single question: do you write and tune production data processing code on Databricks today?

If yes, skip the Associate. If you mostly work in notebooks with managed ingestion and have never investigated why a job costs what it does, the Associate first will be faster overall — the Professional’s 23% code section and 15% optimisation section are unforgiving without that experience.

If you hold the Associate already

Budget three to four weeks rather than six. Concentrate on:

1. Code development — 23%, entirely new emphasis.
2. Cost and performance optimisation — 15%, much deeper than the Associate’s treatment.
3. Data manipulation and data modelling, plus the sharing, federation and CDC material now inside ingestion.
4. Monitoring and alerting — broader than the Associate’s troubleshooting.

Ingestion, governance and transformation you largely know. Verify with practice questions rather than restudying.