The natural assumption is that the Databricks Certified Data Engineer Professional is the Associate exam with harder questions. Same subjects, higher bar.
The published topic weights say otherwise. These are two differently shaped exams, and knowing how they differ changes how you’d prepare for the second one.
The Two Topic Lists Barely Line Up
The Associate, seven topics:
| Topic | Weight |
|---|---|
| Data Transformation and Modeling | 22% |
| Data Ingestion and Loading | 21% |
| Working with Lakeflow Jobs | 16% |
| Governance and Security | 15% |
| Implementing CI/CD | 10% |
| Troubleshooting, Monitoring, and Optimization | 10% |
| Databricks Intelligence Platform | 6% |
The Professional, ten:
| Topic | Weight |
|---|---|
| Developing Code for Data Processing using Python and SQL | 22% |
| Cost & Performance Optimisation | 13% |
| Data Transformation, Cleansing, and Quality | 10% |
| Monitoring and Alerting | 10% |
| Ensuring Data Security and Compliance | 10% |
| Debugging and Deploying | 10% |
| Data Ingestion & Acquisition | 7% |
| Data Governance | 7% |
| Data Modelling | 6% |
| Data Sharing and Federation | 5% |
Read the topic names rather than the percentages and the difference shows up immediately. The Associate names things you do on the platform: ingest, transform, run jobs, set up CI/CD, govern. The Professional names engineering concerns: write the code, control what it costs, debug it, deploy it, secure it.
From Reading Code to Developing It
The largest single block on the Professional is Developing Code for Data Processing using Python and SQL, at 22%. The Associate has no topic of that kind.
The Associate has code on it too, and Databricks is specific about the form: “Data manipulation code in this exam is provided in SQL when possible. In all other cases, code will be in Python.” Provided is the operative word. The Associate shows you code and asks what it does. Its ETL coverage names PySpark directly.
The step up is from reading code to developing it. One exam asks whether you can follow a pipeline someone else wrote; the other gives a fifth of the paper to whether you can write one.
Ingestion makes the same point from the other direction. It’s 21% of the Associate and 7% of the Professional. At Professional level ingestion is assumed, and the marks have moved to what happens after the data lands.
Keep it in proportion. 22% is the biggest topic on the Professional, and the other 78% covers cost, governance, monitoring, security, debugging and deployment. This is a data engineering exam with a coding core.
There’s No Core to Focus On
The Associate concentrates: its top three topics are 59% of the paper. Study transformation, ingestion and jobs properly and you’ve covered most of it.
The Professional flattens out. Its top three are 45%, and six of its ten topics sit at 10% or below. Nothing dominates, and nothing is small enough to skip: 5% of 59 questions is still about three questions, which matters if you’re anywhere near the line.
On the Associate you can triage. On the Professional the weights give you nothing to triage with, so a study plan carried over from the Associate leaves gaps by design.
Format: Longer, Not Faster
- Associate: 45 scored questions, 90 minutes
- Professional: 59 scored questions, 120 minutes
Both work out to almost exactly two minutes per question. The Professional holds that pace over a longer sitting: half an hour more of sustained concentration on denser material.
Both are $200. Both are multiple choice. Both are offered in English, Japanese, Brazilian Portuguese and Korean.
Neither Exam Publishes a Passing Score
Databricks doesn’t state a pass mark on either certification page. Percentages get quoted around both exams, but they don’t come from Databricks, so treat them as unconfirmed rather than as a target to aim at.
The practical response is to stop optimizing for a number nobody has published and prepare until you’re comfortable across every topic. On the Professional that means all ten.
You Can Sit the Professional Without the Associate
Databricks lists no prerequisite for either exam. The Professional’s stated requirement is “None, but related training highly recommended”. You can register for it cold.
Whether you should is a different question. The Associate isn’t a gate. The case for taking it first is that it’s a cheaper way to find out whether your platform knowledge is where you think it is. The case against is $200 and a two-year clock for a credential you may only want as a stepping stone.
If you already work on Databricks daily and write the pipelines yourself, going straight to the Professional is reasonable. If your experience is mostly reading and running other people’s pipelines, the Associate weights sit much closer to your day.
The Two-Year Clock Applies Twice
Both certifications expire after two years, and recertifying means sitting the current version of the exam, not a shorter renewal.
Hold both and you’re maintaining two exams on two clocks at $200 each. That’s worth deciding deliberately rather than discovering when the first renewal notice arrives. Letting the Associate lapse once you hold the Professional is a defensible call, since the higher credential is the one that gets cited.
The other consequence is that “the current version” moves. The Associate guide currently gives 16% to Lakeflow Jobs and 10% to CI/CD, which is a quarter of the paper in two topics. Check that anything you’re studying from is built against the guide as it reads today rather than an earlier revision.
Preparing for the Step Up
Write pipelines rather than read about them. A 22% coding topic isn’t something you can pass on recall.
Then test yourself against the real distribution rather than a general question set. Our Data Engineer Professional practice questions follow the published topic breakdown, and the readiness score reports per topic. A weak area shows up as a number instead of a surprise on exam day. The Data Engineer Associate set is organized the same way against its own seven topics.
If you’re weighing up the Databricks certifications more broadly, the Apache Spark developer exam is a third path. It tests the Spark API itself rather than the platform around it.
