Samsung · Primly Community

Samsung data engineer interview, pipelines and SQL: what they actually care about

ds_dmitri · 4 replies

Just wrapped up a Samsung data engineer interview loop for a role on their mobile platform data team (Plano, TX office). Posting this while it's fresh because I couldn't find much specific info before I went in.

The process was: recruiter screen, a 45-min technical phone screen, then a 4-round onsite (virtual). Two rounds were technical, one was hiring manager, one was a cross-functional stakeholder chat.

Phone screen: Classic SQL, nothing fancy. Window functions, GROUP BY with HAVING, one self-join. I got asked to write a query that finds the second-highest download count per device category. Took me about 8 minutes. Felt fine. They use Spark and Redshift internally, and the recruiter mentioned familiarity with both is "a plus" but not required for senior.

Technical round 1 (pipelines): This is where it got more interesting. They gave me a scenario: you have a near-real-time event stream from Samsung Galaxy devices, 50k events/second. Walk me through how you'd design the ingestion, transformation, and serving layers. I talked through Kafka for ingestion, Flink vs Spark Streaming tradeoffs, a medallion architecture (landing, bronze, silver, gold), and how to handle late-arriving events. They pushed back on my schema choices a few times. Good sign, usually.

Technical round 2 (SQL + data modeling): A mix of query optimization questions (identifying slow queries, indexing strategy, partitioning) and one data modeling exercise around a star schema for a device health monitoring product. I sketched out fact and dimension tables, they asked about slowly changing dimensions. The interviewer knew their stuff.

Behavioral: Pretty standard. Tell me about a pipeline you designed that scaled poorly and what you did. Conflict with a data consumer. A time you had to advocate for tech debt paydown.

Total time from first recruiter reach-out to offer: about 5 weeks. Offer for senior DE came in at $155k base + 10% annual bonus target + RSUs that didn't feel very competitive compared to what I've seen at cloud-native companies. Samsung's comp structure leans more on the salary side than equity.

Overall impression: the technical bar is solid, they genuinely care about the pipeline design layer, not just query writing. Worth prepping your distributed systems fundamentals and be ready to actually talk through tradeoffs.

4 replies

analyst_ana

This is really helpful. Did they ask anything around data quality or monitoring? Like how you'd detect drift or missing data in the pipeline?

de_derek

Yeah, briefly. In the stakeholder round actually, not the technical one. A data consumer asked how I'd alert them if a daily table had 30% fewer rows than expected. I talked through dbt tests, Great Expectations, and a simple row-count anomaly check. They seemed satisfied. Wasn't super deep on that topic.

content_cole

The Kafka + Flink combo showing up at Samsung makes sense. Their IoT device volume is no joke. Curious if they mentioned their internal tooling at all or if it's mostly open source stack.

remote_swe_42

Did you negotiate the RSUs? Samsung notoriously lowballs equity on initial offers. A friend pushed back on the RSU number at the Austin office and got it bumped ~25%.