D.E. Shaw · Primly Community

went through the full DE Shaw loop last fall, here's what actually happened

corp_refugee · 5 replies

applied for a computational role (they call it something like "software development" internally but it's SWE). the recruiter was responsive and clear about what each round was, which i wasn't expecting.

round 1 was a 45-min HackerRank, two problems, one straightforward array manipulation thing and one graph problem that required a non-obvious optimization. i almost misread the constraints on the second one. lesson: read slow.

round 2 was live coding with an engineer, 1 hour. they gave me a problem i'd classify as LeetCode hard but the setting was collaborative. he asked me what i was thinking before i wrote anything, seemed to care about reasoning not just output. no leetcode grinding will prepare you for the "okay now what if the input is 10^9" followup.

round 3 was a systems design conversation. i'd been thinking AWS-scale, they seemed more interested in real tradeoffs at the data layer. lots of "what breaks first" and "how would you know."

final round was behavioral, two interviewers. genuinely interesting conversation about a past project, they pushed pretty hard on decisions i'd made. not gotcha, just curious.

got an offer. took about 9 weeks total from application to verbal. comp was competitive but structured differently than FAANG, more base-heavy.

5 replies

staff_steph

the "what breaks first" framing is so DE Shaw. they really do care about whether you've thought past the happy path. good detail on the timeline, 9 weeks is about right in my experience.

corp_refugee

yeah and they're upfront about it being slow. recruiter told me early on. honestly preferred that to being strung along by firms that say 'we move fast' and then ghost for 3 weeks.

ml_mike

the behavioral round being substantive is accurate. i've heard from people on the quant track that they get a version of this too. it's not just box-checking, they want to see how you think about your own decisions.

newgrad_neil

did you prep specifically for the graph optimization or just happened to know it? asking because i'm not sure how deep to go on advanced graph algos vs. just being solid on the fundamentals

corp_refugee

fundamentals solid plus one solid layer deeper is probably right. i knew dijkstra cold and that was enough to get me to the optimization conversation. i don't think you need to have memorized every variant but you need to be able to reason under pressure.