Nike · Primly Community

Nike data engineer interview, pipelines and SQL: what actually got tested

analyst_ana · 4 replies

Finished the Nike data engineering loop last month. Beaverton-based role, data platforms team supporting digital commerce. Four rounds total, took about five weeks start to finish.

Phone screen was with a recruiter, mostly resume walk-through and a few behavioral questions. Standard stuff. The real thing started with the technical screen: 45 minutes, one SQL problem and one Python scripting question. The SQL was intermediate-hard. They gave me a schema with orders, customers, and product events tables and asked me to write a query returning rolling 7-day revenue by product category with a date filter. Window functions, CTEs, some aggregation. Nothing pathological but definitely not beginner level.

Python question was about parsing a semi-structured JSON log file and extracting error counts by error type. They cared about edge cases: malformed records, nested keys that might be missing, that kind of thing. I wrote clean code with explicit error handling and they seemed to like that.

Onsite was four back-to-back sessions over Zoom (this was remote):

Round 1: Pipeline design. Given a scenario where Nike's DTC sales data and wholesale partner data need to land in a unified analytics layer. Design the ingestion, transformation, and serving layers. I talked through batch vs streaming tradeoffs, brought up Airflow for orchestration, Snowflake as the warehouse. They asked about failure handling and exactly-once semantics, which I was glad I'd thought about.

Round 2: More SQL plus data modeling. Dimensional modeling question, fact vs dimension tables. They asked me to design a schema for tracking athlete endorsement activation data across campaigns. Specific to Nike, which was kind of interesting.

Round 3: Behavioral. Three STAR questions, fairly standard: conflict with a stakeholder, time you shipped something incomplete, how you handle data quality issues discovered in production.

Round 4: Bar raiser-ish. Senior engineer asked about distributed systems concepts, not super deep but they wanted to know I understood partitioning, replication lag, and why idempotency matters in pipelines.

Offer came in around 160-175k base for the Beaverton area, which felt a bit under market for the role scope. Negotiated up slightly. Nike isn't a software-first company so comps are a tick below pure tech.

Bottom line: prep SQL window functions, CTEs, and have a solid pipeline design story. The data modeling question surprised some of my peers who hadn't done it in a while.

4 replies

ux_uma

really useful breakdown. did they test any spark or just python scripting? i've been seeing mixed signals on whether nike uses spark in their stack or if it's mostly dbt/snowflake.

de_derek

mostly dbt/snowflake from what they told me. they mentioned spark existed somewhere in the stack but didn't test on it at all. if you're preparing, focus on SQL and snowflake patterns, not spark internals.

analyst_ana

the dimensional modeling question for athlete endorsement data is fascinating. that sounds like a genuinely tricky schema to design given how a campaign might activate across multiple channels and time windows.

ae_andre

160-175 base in Beaverton is below Seattle/SF but that area has lower COL. still feels like they could do better for a data platforms role. did they offer RSUs?