Revolut Data Scientist Interview Guide

We've seen Revolut lean hard into a very specific profile: someone who can move comfortably between algorithmic coding and applied data work across multiple evaluation stages. Multiple candidates reported that the technical bar was not limited to SQL or product sense — it also included LeetCode-style implementation, Python problem solving, and questions that tested how candidates think through data labeling or KPI measurement. That mix is a strong signal that Revolut wants data scientists who can operate as strong generalists, not just analysts with a notebook. The funnel itself moves from an introductory recruiter call through automated coding assessments, then into technical rounds, a seniority-defining interview, and a culture-fit discussion — so candidates need to sustain that breadth across several distinct conversations over four to six weeks.

A recurring theme is the emphasis on structured reasoning under pressure. Candidates described live coding where they had to explain their thought process as they worked, plus SQL questions involving many joins and open-ended analysis prompts covering product, experimentation, and business impact. More conceptual ML topics also surfaced — assumptions of linear regression, solo ML deployment — which suggests Revolut cares about whether candidates understand how models behave in real systems, not just how to fit them. The non-obvious trap is assuming the interview will stay at the level of dashboards or business metrics; our candidates report that Revolut often pushes into implementation details and model judgment without much warning.

What stands out most is the sheer breadth of evaluation across the funnel. Even when the early conversation felt conversational, the later technical work quickly became dense and multi-disciplinary. The candidates who do best here are the ones who can connect code, SQL, and product logic in one coherent answer — treating them as a unified skill set rather than separate disciplines to compartmentalize.