Spark Developer Associate prerequisites

There are no formal prerequisites for the Databricks Certified Associate Developer for Apache Spark. No certification required, nothing checked at registration. Databricks recommends related training and 6+ months of hands-on practice with the tasks in the exam guide.

What is actually assumed

Python. All code on the exam is Python. You must read PySpark fluently — method chains, arguments, what each call returns. You are not writing code under time pressure, but the reading has to be quick and precise.

Hands-on PySpark. Six months of actually running jobs, not six months of tutorials. The architecture and tuning sections describe symptoms you recognise from having caused them.

SQL. 20% of the exam is Spark SQL. Standard analytical SQL — joins, aggregations, window functions — is assumed.

A cluster to practise on. Preferably with data large enough that partitioning and shuffles have visible consequences. A laptop-scale dataset hides everything the architecture section is about.

What you do not need

• No Scala or Java. This exam is Python only.
• No Databricks platform knowledge. Unusually among Databricks certifications, this one tests Spark rather than the platform. Unity Catalog, Lakeflow Jobs and Delta Lake governance are not covered.
• No distributed systems theory. You need the execution model, not the research literature.
• No streaming production experience. Structured Streaming is 10% and conceptual.

That second point is worth emphasising. If you have used Spark on EMR, Dataproc or on-premises and never touched Databricks, you can still sit and pass this exam. Very little of it is platform-specific.

Ranking your gaps

1. Cannot read PySpark fluently. Blocks the 30% API section and much of the rest.
2. No mental model of the execution engine. Blocks architecture (20%) and makes tuning (10%) incomprehensible.
3. Weak SQL. Blocks 20%.
4. Never seen a real performance problem. Concentrated in tuning, and hard to learn without a cluster and real data.
5. No streaming exposure. Only 10% and conceptual — the most forgiving gap.

If you are short of the six months

Build something. Take a dataset of a few hundred million rows and:

• Write transformations with joins, aggregations and window functions.
• Open the Spark UI after every run and look at stages, tasks and shuffle volume.
• Create a skewed key deliberately and watch what happens.
• Broadcast a small table and compare the plan before and after.
• Write the same logic in SQL and in the DataFrame API and compare the physical plans.

A few focused weekends of that substitutes reasonably for months of incidental exposure, because it targets exactly what the exam tests.

The portability advantage

Unlike the other Databricks certifications, what you learn here applies wherever Spark runs. If you are unsure whether your next role will use Databricks, this is the credential in their range whose knowledge travels — which also makes it a reasonable first Databricks certification for someone hedging.

Before you book

1. Read a twenty-line PySpark script and explain what each line does.
2. Name three wide transformations and three narrow ones.
3. Open the Spark UI and find the shuffle volume for a job.
4. Write the latest-row-per-group pattern with a window function.
5. Try the free sample questions.

If step 2 is uncertain, start with week 1 of the study plan — the architecture section makes everything else easier.