UBS Data Scientist interviews typically start with an online assessment, move into a statistics-heavy technical round on regression and regularization, then shift to business or behavioral discussion and a final panel or HR conversation; one candidate reported the full process took about three weeks.
$137K
Avg. Base Comp
$202K
Avg. Total Comp
3-4 rounds
Typical Rounds
3 weeks
Process Length
UBS Data Scientist interviews, based on candidate reports, typically begin with an online assessment before any live conversation. Candidates describe two different assessment styles — a technical-and-cultural screen that gates further rounds, and a separate version built around logic and small math problems, including IQ-like spatial reasoning. From there, the process leans heavily on applied statistics and past project depth rather than algorithmic coding drills. Expect to explain regularization methods like ridge regression and lasso, justify why you'd use k-fold cross-validation, and defend specific technical choices from your own projects — one candidate was asked directly why they chose a NoSQL approach over SQL for a past build.
Later rounds shift toward business judgment and team fit. One candidate reported discussing how they'd frame a problem from a business angle, and that interview touched on a domain challenge: using machine learning for reconciliation work. Final stages vary — one candidate described a dedicated HR conversation about growth and expectations, while another completed a 90-minute interview covering both technical and behavioral ground with two team members. One candidate reported the process spanning about three weeks across multiple rounds.
Communication after interviews is worth planning for. One candidate described strong, professional technical interviewers but then silence and unresponsiveness from recruiting, even though the role reportedly stayed open more than 45 days after the interview. Another candidate described an organized process, so this may not be universal, but it's reasonable to follow up proactively if you don't hear back.
Synthesized from 3 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 Ubs process.
I initially had an online assessment with logic and small math problems, then a video interview, and then a 90-minute interview with two members of the team. They asked a mix of technical and behavioral questions.
In the online assessment, there were some IQ-like spatial problems.
The technical questions were questions like:
In the technical deep dive, I mostly had to talk about my projects, for example, "Why did you decide to go for a NoSQL instead of a SQL approach?"
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 Ubs
Which model would perform better and why?
| Question | |
|---|---|
| Hurdles In Data Projects | |
| Lasso vs Ridge | |
| Data Preparation for Imbalanced Data | |
| SARIMA in Retail Forecasting | |
| Shortest Path Algorithms | |
| Decision Tree Evaluation | |
| Stakeholder Communication | |
| Company Acquisition Choice | |
| Your Strengths and Weaknesses | |
| Martingale Strategy | |
| Student Tests | |
| Statistically Significant Test | |
| Adam Optimizer | |
| 2nd Highest Salary | |
| Employee Salaries | |
| Bagging vs Boosting | |
| Random SQL Sample | |
| P-value to a Layman | |
| Size of Joins | |
| WAU vs Open Rates | |
| Random Forest Explanation | |
| Type-ahead Search | |
| Cyclic Detection | |
| Scalped Ticket | |
| Three Zebras | |
| Missing Housing Data | |
| Precision and Recall | |
| Assumptions of Linear Regression | |
| Integer String Addition |
Synthesized from candidate reports. Individual experiences may vary.
Candidates report starting with an online assessment before any live interviews. One candidate described a technical and cultural fit assessment that had to be passed to move forward, while another described an assessment built around logic and small math problems, including IQ-like spatial reasoning questions. Expect quantitative reasoning and short technical prompts rather than an open-ended coding challenge at this stage.
The core technical round centers on applied statistics rather than algorithmic coding. Candidates report being asked to explain regularization methods like ridge and lasso regression, discuss why k-fold cross-validation is used, and walk through past projects in detail, including specific choices such as why they picked a NoSQL over a SQL approach. Interviewers reportedly probe the reasoning behind technical decisions, not just the final result.
A separate round shifts toward behavioral and business judgment. One candidate reported discussing how they would approach problems from a business perspective, how they would fit the team culture, and a domain-specific challenge involving reconciliation using machine learning and AI. This round appears to test communication and business framing alongside technical depth, though this level of detail comes from a single account.
Later stages can take different shapes. One candidate described a 90-minute session with two team members mixing technical and behavioral questions, while another described a separate closing HR discussion focused on role expectations and growth. Candidates should be ready for either a panel-style technical/behavioral mix or a more HR-oriented final conversation, since the reported formats differ.
One candidate reported a long, unexplained silence after the technical round, with the position reportedly still open more than 45 days later and no status update despite repeated follow-up attempts. Another applicant described an organized process, so this may not be universal, but it's worth planning for possible delays and following up proactively.
