EY · Primly Community

EY data engineer interview: pipelines and SQL were basically the whole thing

de_derek · 4 replies

Went through the EY data engineer interview loop last month for a role in their Technology Consulting practice. Sharing notes because I couldn't find much specific info before I went in.

Total rounds: four. Phone screen with a recruiter (standard 20 min, background + availability), then a technical screen with someone from the data engineering team, then two final-round interviews on the same day (one technical, one behavioral/case-mix).

The SQL was real. Not "write a SELECT" SQL. More like: given this messy schema, write a query that handles NULLs correctly, uses window functions to get running totals, and explain why your approach is better than a subquery. One question gave me a slowly changing dimension scenario and asked how I'd handle it in a query before diving into the architecture side. Intermediate-to-advanced range. Know PARTITION BY, LEAD/LAG, RANK vs DENSE_RANK.

Pipeline questions were more design-oriented. They asked me to walk through how I'd build an ELT pipeline for a client that's ingesting data from 10+ source systems with different schemas. I talked about orchestration (they seemed fine with Airflow references), idempotency, schema evolution, monitoring. No right answer, they wanted to see that I thought about failure modes.

They also asked about Spark. Not deep Spark tuning, more like: do you know the mental model, have you used it in anger. I had real experience so it went fine, but I wouldn't fake it.

Behavioral side was fairly standard EY stuff: client impact, dealing with ambiguity, a time you had to simplify technical concepts for a non-technical stakeholder. They care a lot about that last one because the work is consulting-adjacent and you will be explaining data pipelines to clients who don't know what a DAG is.

Timeline was about 3 weeks start to finish. Offer came ~5 business days after the final round.

Happy to answer specifics if anyone's in the same loop.

4 replies

ds_dmitri

The SCD scenario in SQL is a great data point. Did they want you to solve it in pure SQL or did they pivot to asking how you'd design the table/warehouse layer to handle it? I always get that question mid-answer and it shifts the whole conversation.

de_derek

Both, kind of sequentially. Started with SQL (they wanted to see I could write the merge logic), then opened up to 'ok so how would you actually model this in a DWH.' I mentioned Type 2 SCDs with effective dates and they seemed happy with that direction. Didn't get super deep on the warehouse modeling but it's clearly fair game.

analyst_ana

Was this a specific industry practice or general tech consulting? I've been applying to a few EY data roles and the JDs vary a lot between FSO, health, government. Wondering if the interview format differs by practice.

visa_vik

Thanks for this. I'm targeting EY for H1B sponsorship reasons and the consulting track seems more reliable than pure product companies right now. Did the recruiter bring up visa support at all or did you have to ask?