Free Databricks Data Engineer Associate questions
Try these before reading the answers. The mix follows the weighting, so ingestion and transformation appear most.
Question 1. A nightly job reprocesses an entire cloud storage location even though only a small number of new files arrive each day. Cost and runtime are both rising. What should the engineer implement?
- A. A larger cluster to reduce runtime
- B. Incremental ingestion that processes only newly arrived files
- C. Convert the source files to a different format
- D. Run the job weekly instead of nightly
Show answer
Answer: B
Incremental ingestion that processes only newly arrived files addresses both cost and runtime directly, since the problem is reprocessing unchanged data. A larger cluster makes the same wasteful work faster and more expensive, converting formats does not reduce what is read, and running less often increases data latency without fixing the inefficiency.
Question 2. A transformation job is occasionally rerun after a failure, and this has produced duplicate rows in the target table. What property is missing?
- A. Correct partitioning on the target table
- B. Sufficient cluster capacity
- C. Idempotency — for example, a merge on a unique key rather than an append
- D. Delta Lake time travel enabled
Show answer
Answer: C
A pipeline that produces different results when rerun is not idempotent, and a merge keyed on a unique identifier makes reruns safe. Partitioning affects layout, a larger cluster affects speed, and time travel lets you read earlier versions without preventing duplication.
Question 3. An analyst must be able to read one table but must not see a column containing national identifiers. Which capability provides this?
- A. Unity Catalog column-level access control or masking
- B. A separate copy of the table without that column
- C. A job parameter that excludes the column
- D. A cluster policy restricting the analyst’s compute
Show answer
Answer: A
Unity Catalog governs access at table and column level, so a column can be masked or withheld while the rest of the table remains readable. A separate copy duplicates data and creates drift, a job parameter does not control access, and cluster policies constrain compute rather than data visibility.
Question 4. A Lakeflow Job has three tasks. Task C must run only if both A and B succeed, and the whole job should stop if A fails. How is this expressed?
- A. Three separate jobs scheduled a few minutes apart
- B. One task containing all three steps in sequence
- C. Running the tasks manually in order each day
- D. Task dependencies in the job, with C depending on both A and B
Show answer
Answer: D
Task dependencies in the job definition express that C depends on both A and B, and a failed upstream task prevents dependent tasks from running. Three separate scheduled jobs cannot enforce ordering, a single task loses the structure, and manual sequencing defeats the purpose of orchestration.
Question 5. Queries against a Delta table have become slow. Investigation shows the table consists of hundreds of thousands of very small files. What is the appropriate remedy?
- A. Increase the cluster size
- B. Compact the small files into fewer, larger files
- C. Add more partition columns to the table
- D. Convert the table to a non-Delta format
Show answer
Answer: B
The small file problem is resolved by compacting data into fewer, larger files so queries read less overhead, which is what optimising file layout does. A larger cluster masks the cost, more partitions usually worsens fragmentation, and converting to a non-Delta format loses the platform’s capabilities without addressing layout.
How did you do?
Notice that every question describes a symptom and asks for the cause or the remedy. That is how this exam is written — it assumes you have run pipelines and seen them misbehave.
If the ingestion and transformation questions felt hardest, that matches the weighting: they are 43% between them. See weeks 1 and 2 of the study plan.