Data Engineer interview questions
Grouped by interview stage, with what a strong answer actually sounds like — the part most question lists leave out.
When interviewing a data engineer, structure questions by stage and assess against the role's actual requirements. Data engineers build the pipelines and models that make data usable. The questions below cover screening through final round, each with what a strong answer sounds like and the warning signs of a weak one.
How to use these
Pick four to six for a 45-minute interview — not all 7. A well-followed-up question takes six to eight minutes, and the information is almost always in the follow-up rather than the first answer. Ask the same core questions of every candidate so you can actually compare them; vary the follow-ups.
Screening questions
Tell me about a pipeline you built. What broke, and how did you find out?
Why ask it: Every pipeline breaks. How they learned about it reveals their quality practice.
A strong answer: Found out from monitoring, not from a stakeholder. Explains the check that caught it.
A weak answer: Found out when someone complained the dashboard looked wrong.
How do you test data?
Why ask it: The distinguishing habit between data engineers and people who write SQL.
A strong answer: Freshness, uniqueness, referential integrity, volume anomalies, tested in the pipeline.
A weak answer: Eyeballs the output.
Technical questions
Two dashboards show different numbers for the same metric. How do you resolve it?
Why ask it: The most common and most trust-destroying data problem.
A strong answer: Traces both to source, finds the definitional difference, then fixes the definition centrally.
A weak answer: Picks whichever looks right.
How do you handle late-arriving data?
Why ask it: Separates people who have run pipelines in production from those who have built them once.
A strong answer: Discusses windowing, reprocessing and idempotent loads.
A weak answer: Has not encountered it.
When would you denormalise?
Why ask it: Tests modelling judgement rather than recall.
A strong answer: For query performance and analyst usability, accepting the maintenance cost knowingly.
A weak answer: Always or never, with no reasoning.
Final questions
How do you work with analysts who need something urgently?
Why ask it: Data teams live under constant urgent requests.
A strong answer: Distinguishes genuinely urgent from routine; builds self-service for repeats.
A weak answer: Drops everything each time, or refuses all interruptions.
How do you decide a dataset is trustworthy enough to publish?
Why ask it: Surfaces their standard, which varies enormously.
A strong answer: Tests passing, lineage documented, owner named, known caveats written down.
A weak answer: Publishes and waits for complaints.
Red flags for this role
- No data quality testing at all
- Cannot explain a model they built
- Treats stakeholder trust as someone else's problem
What a data engineer actually does
Data engineers build the pipelines and models that make data usable. The work is judged on whether the numbers people rely on are correct, timely and explainable.
See the full data engineer job description, including how to assess each requirement.
Generate questions specific to your role — and that account for what previous interviewers already asked — with the free interview question generator.