Two Sigma · Primly Community

Two Sigma data engineer interview: pipelines and SQL, what they actually focus on

analyst_ana · 4 replies

Just finished my Two Sigma data engineer loop last week. Sharing because the internet has way more SWE content about them than DE content and I had to piece this together from scraps.

First: they are very much a quant shop, but the DE role I interviewed for (their Systematic Trading infra team) was genuine engineering work. Not just writing queries for quants. So don't be scared off by the math-heavy reputation if you're coming from a traditional DE background.

The loop (NYC, 2026): Phone screen with recruiter, then 45min with an engineer. Mostly SQL. They gave me a schema for something like an order book with timestamps and asked me to write queries to compute rolling averages, identify gaps in time series, and do a self-join to find matching events. Nothing exotic but a few tricks in there. Window functions were essential. Virtual onsite: 4 rounds. One pure coding (Python, data manipulation, basically LeetCode medium with a Pandas flavor). One system design focused on pipeline architecture: I had to design an ingestion layer for market event streams. They cared a lot about latency guarantees, exactly-once semantics, and how you'd handle schema evolution. One SQL deep dive, more complex than the phone screen. One behavioral/cross-functional round.

What they pushed on in system design: They kept asking "what breaks first under load" and "what do you sacrifice when you can't have all three of consistency, availability, latency." This felt very quant-informed. They want people who have a mental model for tradeoffs, not people who just know the buzzwords.

For SQL, practice recursive CTEs and window functions until they're automatic. Also know the difference between RANK, DENSERANK, ROWNUMBER -- they asked about this explicitly.

Behavioral: Pretty standard STAR stuff. They wanted to hear about cross-functional work and times I pushed back on stakeholders. My interviewers were sharp and engaged, not phone-it-in energy.

Total timeline was about 5 weeks from recruiter reach-out to verbal offer.

Happy to answer specifics about the system design round if anyone's heading into it.

4 replies

analyst_ana

this is exactly what i needed, thank you. was the sql during the onsite timed or more of a conversation? like did they give you a text editor and watch you type or was it more whiteboard-style?

mobile_mara

it was a shared coding environment (coderpad i think). they can see you type in real time. not whiteboard. i talked through my thinking as i went and they'd occasionally drop hints if i went sideways. felt collaborative enough, not adversarial.

sre_sol

the exactly-once semantics question is a classic at quant shops. they want to know if you understand idempotency keys and two-phase commit vs. outbox pattern, not just 'use kafka.' did you end up going kafka or something else in your design?

remote_swe_42

were the interviewers from the DE team specifically or was it a mixed panel? trying to gauge how deep the domain questions go vs. generic eng.