Robinhood · Primly Community

Went through the full Robinhood loop last month. Here's what actually happened.

backend_bekah · 5 replies

Applied in late April, got a recruiter screen within a week. Fairly standard: role fit, location (they are hybrid in Menlo Park and NYC, remote for some teams), compensation expectations. Recruiter was organized and sent a prep guide right away, which I appreciated.

Technical phone screen was 45 minutes on CoderPad. The interviewer gave me a medium LC problem, graph traversal variant. I finished with time left, we talked through complexity, then he asked me to optimize it. Felt conversational.

Virtual onsite had 5 rounds back to back over two days. Coding rounds: one array/hashmap problem, one binary search variant that felt designed to trip you up on edge cases. I got dinged on the second one because I missed an off-by-one in a time-constraint scenario. Not going to pretend otherwise.

System design was interesting: design a notification system for trade confirmations that has to handle spikes (like a meme stock day). They were really focused on queue depth, backpressure, and what happens when downstream services fall behind. I talked through Kafka, partitioning strategy, and retry logic. Got good engagement from the interviewer.

Behavioral round was the longest and felt genuinely probing. Questions about a time I had to push back on a decision, a time I shipped something that broke trust, how I communicate technical risk to non-engineers. Not softballs.

Got an offer about 10 days after the onsite. No lowball. Overall the process was well-run.

5 replies

sre_sol

the notification system design prompt is evergreen for fintech. they also ask a version of it at Stripe and Cash App. the backpressure angle is the right one to push on. what did you use for the retry logic? exponential backoff or something more custom?

backend_bekah

exponential backoff with jitter and a dead-letter queue after N retries. they seemed happy with that. one interviewer asked what happens if the DLQ itself fills up, which caught me off guard a little. i said alert on queue depth and degrade gracefully. probably not the most elegant answer but they moved on.

market_realist

appreciate the honest writeup. the off-by-one getting flagged is real, they do care about correctness not just approach. did the behavioral round feel like a filter or more like a conversation?

corp_refugee

the 'time you shipped something that broke trust' question is doing a lot of work. that's not a question a company asks when they have a healthy relationship with failure. just noting it.

backend_bekah

fair read. i took it as them wanting to see self-awareness, not accountability theater. the follow-up they cared about was what i changed afterward. but your interpretation is also valid.