
A candidate for a Brex senior data scientist role reports a recruiter call, a project discussion plus statistics coding round, and a two-day onsite with four interviews, where the first two were gating.
$196K
Avg. Base Comp
$325K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
The one detailed report we have, from a candidate interviewing for a senior data scientist role at Brex, describes a process that felt more like an exam than a conversation. After a recruiter screen, the second round paired a walkthrough of past project work with a statistics coding round. The onsite was split across two days with four interviews, and the candidate says the first two were gating.
Statistics fundamentals and algorithmic coding stand out in this account. The candidate describes the stats material as basics that still needed a refresh, nothing exotic. The coding was described as close to LeetCode, with medium-to-hard algorithmic problems that had little to do with day-to-day data science, even though the recruiter said it would not be traditional LeetCode. The candidate also met ML fundamentals, product-specific data science applications and behavioral questions. They were asked what four things they would ask when taking over someone else's model.
This candidate did not receive an offer, so the account says little about what separates a successful loop. A sensible plan is to rehearse your project story, refresh core statistics, and practice timed algorithm problems. Also prepare a clear framework for inheriting, debugging and taking ownership of a model.
Synthesized from 1 candidate report by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Brex process.
I got approached by a recruiter and went through their full interview process. The first round was just a recruiter call to screen fit and experience. The second round had two parts that happened back-to-back: a project discussion where I walked through work I'd done previously, and then a statistics coding round. The recruiter said it wouldn't be traditional LeetCode, but honestly it felt pretty close to that — medium to hard algorithmic problems that didn't have much to do with data science work. Then came the onsite, split across two days with four interviews total. The first two were gating — if you didn't do well, you didn't move on to the last two. Those covered coding (again, not LeetCode in theory but pretty similar), ML fundamentals, data science applications specific to their product, and behavioral questions. One interviewer asked me to walk through four things I'd ask when taking over someone else's model, which was actually a solid question about ownership and debugging. The whole thing felt like an exam more than a conversation. I found the stats questions required brushing up on basics — nothing exotic, just making sure you knew your fundamentals. The coding was the wildcard though. Some interviewers seemed great and calibrated, others felt like they were just checking if you'd seen the problem before. The recruiter was responsive and gave feedback after each round, which was helpful for understanding what they were looking for, even if the feedback didn't always map to what actually helped.
Prep tip from this candidate
Focus hard on stats fundamentals (distributions, hypothesis testing, confidence intervals) since that's an explicit gate. The coding round heavily favors LeetCode-style problems despite their claim otherwise — be ready for medium-to-hard algorithmic problems, not just data science-specific coding. Prepare a few model handoff and debugging frameworks you can articulate clearly.
Share your own interview experience to unlock all reports, or subscribe for full access.
Sourced from candidate reports and verified by our team.
Topics based on recent interview experiences.
Featured question at Brex
Justify an array of words given an arbitrary line width.
| Question | |
|---|---|
| Hurdles In Data Projects | |
| Inherited Model Evaluation | |
| Fixed Length Arrays: Addition | |
| Client Solution Pushback | |
| 2nd Highest Salary | |
| Empty Neighborhoods | |
| Rolling Bank Transactions | |
| Closest SAT Scores | |
| Employee Salaries | |
| Comments Histogram | |
| Top Three Salaries | |
| Last Transaction | |
| Subscription Overlap | |
| Button AB Test | |
| Cumulative Distribution | |
| Monthly Customer Report | |
| Payments Received | |
| Top 5 Turnover Risk | |
| Paired Products | |
| P-value to a Layman | |
| Merge Sorted Lists | |
| Total Spent on Products | |
| Find the First Non-Repeating Character in a String | |
| Experiment Validity | |
| Top 3 Users | |
| Swipe Precision | |
| Testing Price Increase | |
| Bank Fraud Model | |
| String Shift |
Synthesized from candidate reports. Individual experiences may vary.
The candidate was approached by a recruiter, and the first round was a call to screen fit and experience. Expect a conversation about your background and role fit. The candidate also says the recruiter was responsive and gave feedback after each round.
In the second round, the candidate reports a project discussion where they walked through work they had done previously. It ran back-to-back with the statistics coding portion. Prepare to explain your decisions, results and ownership clearly, since this may set up later questions.
The candidate reports this round required brushing up on statistics basics and felt close to LeetCode, with medium-to-hard algorithmic problems despite the recruiter saying otherwise. Preparing for both statistics concepts and general coding is sensible, though one report cannot show how the round is run for every candidate.
The candidate describes an onsite split across two days with four interviews, and says the first two were gating for the last two. Topics reported include coding, ML fundamentals, product-specific data science applications and behavioral questions. One interviewer asked for four things they would ask when taking over someone else's model.