Santander Data Scientist interview typically runs 2 rounds: hiring manager, technical discussion. Timeline is about 1-2 weeks, and the process is compliance-focused and interview-heavy.
$115K
Avg. Base Comp
$125K
Avg. Total Comp
3
Typical Rounds
1-2 weeks
Process Length
We’ve seen Santander lean hard into interpretability over sophistication, especially for data science roles tied to credit and lending. In the candidate experience we reviewed, the conversation quickly moved from background and project history into why a model behaves the way it does, not just whether it performs well. That makes sense for a bank operating under heavy compliance pressure: a strong NLP or deep learning background may get attention, but it doesn’t replace the need to defend simpler models in a regulated setting.
A recurring theme is that the team seems to probe whether candidates can connect modeling choices to the business context. One candidate was pressed on BERT architecture because of prior NLP work, but the discussion pivoted just as quickly to linear and logistic regression, regularization, and the meaning of each term in the equation. That tells us Santander is looking for people who can explain why a white-box approach is preferred and how they would handle practical issues like missing values with a mix of statistical methods and domain input.
The non-obvious separator here is depth, not breadth. Our candidate felt comfortable on the BERT and imputation portions, but struggled when the interviewers dug into optimization and the mechanics of linear models. That’s a useful signal: Santander appears to reward candidates who can move past “I’ve used this before” into a more rigorous, first-principles understanding of the tools they’ve applied.
Synthetized from 1 candidates 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 Santander process.
I interviewed at Santander for a Senior Associate role in Data Science within their Credit and Lending division. The process started with a hiring manager round that was mostly experience-based — walking through my resume, discussing my background, and getting an overview of what the team and vertical does. Because my experience is inherently technical, the conversation naturally drifted into technical territory, but it wasn't a formal technical screen.
The second round was a proper technical discussion, running about 45–50 minutes. The interviewers knew I had NLP experience and a master's thesis on BERT, so they opened by going deep on the BERT architecture — specifically how the attention mechanism works. From there, they pivoted to linear modeling, which made sense given the credit and lending context. Banks like Santander operate under heavy compliance obligations, so they favor white-box, interpretable models over black-box approaches like gradient boosting or deep learning. They asked about how linear regression and logistic regression work, how to regularize each using L1 and L2, what each term in the logistic regression equation means, and how you'd optimize those components. They also asked about null value handling — specifically what methodologies you'd use, whether you'd lean on subject matter expertise from the business, or apply statistical imputation methods like mean/median/mode, or go with regression-based imputation.
I felt confident on the BERT and NLP portions, and the data imputation discussion went well. Where I struggled was on the interpretability and optimization side of linear and logistic regression — I had used regression in various projects but hadn't focused deeply on that layer of understanding. That gap likely cost me the round, as there was a third round (a two-hour virtual onsite) that I wasn't moved forward to.
Prep tip from this candidate
Santander's credit and lending team explicitly tests interpretability of linear models — be ready to explain every term in the logistic regression equation, how L1 vs. L2 regularization affects feature selection, and how you'd optimize each. Also prepare a structured answer on null value imputation strategies, covering business-driven decisions, statistical methods (mean/median/mode), and regression-based approaches.
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 Santander
Given a list of locations that your trucks are stored at, return the top location for each model of truck (Mercedes or BMW).
| Question | |
|---|---|
| 2nd Highest Salary | |
| Empty Neighborhoods | |
| Merge Sorted Lists | |
| Rolling Bank Transactions | |
| Comments Histogram | |
| Employee Salaries | |
| Closest SAT Scores | |
| Top Three Salaries | |
| Subscription Overlap | |
| Experiment Validity | |
| Find the Missing Number | |
| Cumulative Distribution | |
| Compute Deviation | |
| Maximum Profit | |
| Prime to N | |
| String Shift | |
| Bagging vs Boosting | |
| Last Transaction | |
| 500 Cards | |
| Session Difference | |
| Rain in N Days | |
| Random SQL Sample | |
| Alphabet Sum | |
| Paired Products | |
| Bank Fraud Model | |
| Rectangle Overlap | |
| Swipe Precision | |
| Hurdles In Data Projects | |
| Unique Work Days |
Synthesized from candidate reports. Individual experiences may vary.
The process starts with a hiring manager interview that is mostly experience-based. Candidates walk through their resume, discuss their background, and get an overview of the team and the Credit and Lending vertical. The conversation can drift into technical territory, but it is not described as a formal technical screen.
This round is a deeper technical discussion focused on the candidate's background and the needs of the role. In the reported experience, interviewers asked about BERT and attention mechanisms, then moved into linear and logistic regression, L1/L2 regularization, interpretability, optimization, and null value handling/imputation methods. The emphasis reflects Santander's preference for white-box, compliant models in credit and lending.
The candidate reported that there was a third round described as a two-hour virtual onsite, though they were not advanced to it. No additional details were provided about the format or topics covered.