title: "Data Engineer Interview Questions at Uber (2026 Guide)" slug: "uber-data-engineer-interview-questions" description: "Uber data engineer interview: rounds, SQL, Spark, data modeling and pipeline design questions, plus FAQs for 2026."
Data Engineer Interview Questions at Uber (2026 Guide)
Uber's data engineer interview runs 4-5 rounds: recruiter screen, a technical screen split between SQL and coding, then an onsite with a deep SQL round, a data modeling round, a pipeline/system design round, and a behavioral round. Difficulty is moderate-to-hard. Uber built and open-sourced serious data infrastructure (Hudi, and heavy Spark/Kafka/Presto/Flink usage), so pipeline design questions go deep on scale, late data, and exactly-once semantics. Expect 3-5 weeks.
Uber Data Engineer Interview at a Glance
| Stage | Format | Typical Duration | Focus |
|---|---|---|---|
| Recruiter screen | Phone | 30 min | Stack fit, level |
| Technical screen | Video | 60 min | SQL + Python problem |
| Onsite: SQL deep dive | Live | 45-60 min | Window functions, optimization |
| Onsite: data modeling | Live | 45-60 min | Star schemas, SCDs, trip-domain models |
| Onsite: pipeline design | Video/whiteboard | 60 min | Batch + streaming architecture |
| Behavioral | Conversation | 45 min | Ownership, cross-team work |
Estimates from candidate reports. Stack to know: Spark, Kafka, Flink, Hive/Presto, Airflow-style orchestration, Hudi/lakehouse concepts.
Real-Style Interview Questions by Round
SQL
- For each city, find the top 3 drivers by completed trips per month (window functions).
- Compute session-level metrics from raw event logs with gaps over 30 minutes defining a new session.
- This query scans 2 TB and takes an hour. Partitioning, clustering, rewrite — optimize it.
Coding
- In Python, merge late-arriving events into an ordered stream with bounded memory.
- Parse malformed JSON logs, quarantine bad records, and report error rates.
- Implement a hash-based data-skew detector for join keys.
Data modeling
- Model the trip lifecycle (request, match, pickup, dropoff, payment) for analytics. Facts, dimensions, grain?
- How do you handle a slowly changing driver dimension (SCD2) when drivers change vehicles?
- One events table gets 50B rows/day. Partitioning and retention strategy?
Pipeline / system design
- Design the pipeline that powers city-level supply-demand dashboards with under 1 minute of lag.
- Design ingestion for GPS pings from millions of devices — dedup, ordering, and late data.
- Batch job backfills 90 days of data while streaming continues. How do you avoid double counting?
- Exactly-once from Kafka to the lakehouse: what actually guarantees it?
Behavioral
- Tell me about a data quality incident you caught (or missed). What changed?
- Describe negotiating schema changes with a producer team that didn't care about your pipelines.
FAQs
How hard is the Uber data engineer interview? Moderate-to-hard. SQL must be fast; the separator is streaming design — late data, ordering, exactly-once.
Does Uber ask LeetCode for data engineering? Light-to-medium coding, usually data manipulation in Python rather than pure algorithms.
What tools should I know? Spark, Kafka, Flink, Presto/Hive, and lakehouse table formats (Uber created Apache Hudi).
Is there a take-home? Usually no; rounds are live (candidate reports vary by team).
How long is the process? Typically 3-5 weeks (estimate).
<script type="application/ld+json"> {"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How hard is the Uber data engineer interview?","acceptedAnswer":{"@type":"Answer","text":"Moderate-to-hard. Fast SQL is required; streaming design with late data and exactly-once semantics separates candidates."}},{"@type":"Question","name":"Does Uber ask LeetCode questions for data engineering?","acceptedAnswer":{"@type":"Answer","text":"Light-to-medium coding, usually Python data manipulation rather than pure algorithms."}},{"@type":"Question","name":"What tools should I know for Uber data engineering?","acceptedAnswer":{"@type":"Answer","text":"Spark, Kafka, Flink, Presto/Hive, and lakehouse formats — Uber created Apache Hudi."}},{"@type":"Question","name":"How long does Uber's data engineer interview take?","acceptedAnswer":{"@type":"Answer","text":"Typically 3-5 weeks from screen to offer."}}]} </script>Related Guides
- Uber Data Scientist Interview Questions
- Uber Software Engineer Interview Questions
- Uber Senior Software Engineer Interview Questions
- Airbnb Data Scientist Interview Questions
Stream the Right Answer, Zero Lag
InterviewBoost.ai's Live Interview Assist delivers suggested answers in 0.3 seconds on Zoom, Teams, or Meet. Drill Spark, Kafka, and modeling questions in AI mock interviews before the real loop. 4.9/5 from 50K+ job seekers. Start your free one-week trial — then $99/mo.