
Two candidates describe the Braze Data Scientist interview as an online assessment followed by technical rounds on ML theory, coding or data engineering, and a live data science project. Both reported no offer.
$168K
Avg. Base Comp
$202K
Avg. Total Comp
Not reported
Typical Rounds
Not reported
Process Length
Two candidates describe a Braze Data Scientist process that starts with an online assessment and then moves through several technical sessions. One mentions an initial call and a HackerRank test. The other describes an online assessment followed by three sessions: Python and data engineering with behavioral questions, ML knowledge, and a one-hour live data science project.
The ML theory content was specific. One candidate was asked to break down an automated daily workflow for a reinforcement learning project recommending renewal offers with a gradient boosting classifier. The same candidate had to set up a hypothesis test for an upsell experiment with a 1.6% baseline conversion rate and 25% expected uplift. They were also asked about bias versus variance, gradient descent versus gradient boosting, and reinforcement learning versus supervised learning. That points to practicing how you would explain A/B test design, significance and model training choices out loud.
The other candidate's live project was a customer offer dataset analyzed within an hour. It covered exploratory analysis, class imbalance, invalid values, a logistic regression baseline, evaluation with ROC-AUC and PR-AUC, and coefficient interpretation for a business audience. One candidate also mentioned pandas date-time edge cases in the data engineering round.
Both candidates received rejections. One felt the bar leaned toward textbook definitions. The other had positive feedback in each session but noted that the selected candidate had a lifecycle marketing background, and that feedback took over a week despite a stated 48-hour window. With only two reports, treat these as individual experiences. Ask your recruiter what the team weighs beyond the posted skills.
Synthesized from 2 candidate reports by our editorial team.
Had an interview recently?
Share your experience. Unlock the full guide.
Real interview reports from people who went through the Braze process.
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 Braze
Instagram third party messaging
| Question | |
|---|---|
| Statistically Significant Test | |
| Offer Matching API Design | |
| 2nd Highest Salary | |
| Empty Neighborhoods | |
| Upsell Transactions | |
| Closest SAT Scores | |
| Top Three Salaries | |
| Rolling Bank Transactions | |
| First to Six | |
| Comments Histogram | |
| Merge Sorted Lists | |
| Longest Streak Users | |
| Last Transaction | |
| Subscription Overlap | |
| Customer Orders | |
| Paired Products | |
| Top 3 Users | |
| Random SQL Sample | |
| Raining in Seattle | |
| Employee Salaries (ETL Error) | |
| Size of Joins | |
| Find the Missing Number | |
| Fair Coin | |
| Monthly Customer Report | |
| Impression Reach | |
| Manager Team Sizes | |
| Largest Salary by Department | |
| Scrambled Tickets | |
| Experiment Validity |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report an online assessment early in the process. One also mentions an initial call and a HackerRank test beforehand. The reports do not describe the assessment content in detail, so prepare broadly for Python and data-handling problems.
One candidate describes a round of Python and data engineering with behavioral questions, and another mentions pandas manipulation with date-time edge cases. Practice date-time handling in pandas and prepare a few behavioral examples, though exact questions may differ.
Candidates report conceptual ML questions. One described an automated reinforcement learning workflow with gradient boosting, hypothesis testing for a low-conversion upsell test, bias versus variance, and gradient descent versus gradient boosting. Expect to explain definitions and reasoning clearly.
One candidate reports a live classification problem with about 15 features. Another describes a one-hour project on a customer offer dataset covering EDA, imbalance, logistic regression, ROC-AUC and PR-AUC, and business interpretation. Practice finishing an end-to-end analysis within a fixed time.
Both candidates report no offer. One describes a generic rejection after several rounds, and the other reports feedback taking longer than the stated 48 hours. Timing may vary, so ask the recruiter about expected next steps and evaluation criteria.