Data Scientist Interview Questions UK (2026)

UK Data Scientist interviews run longer and test more layers than Data Analyst interviews, typically four to six stages across two to four weeks. The core is consistent across most UK employers in 2026: statistics and probability, machine learning judgement, SQL under time pressure, and a business case round that checks whether you can connect a model to a decision, not just a metric. This guide covers the questions that come up repeatedly, grouped by stage.

If you have not built the portfolio these interviews test yet, pair this with our Data Scientist portfolio guide.

Round 1: The Recruiter Screen (2 questions)

1. Walk me through a project where you moved from a business question to a model.

The interviewer is checking for structure, not technical depth yet. A strong answer names the business question first, then the data, then the method, then what happened when the model met reality. Weak answers start with the algorithm.

2. Why Data Science, and why this company specifically?

Generic answers (“I love machine learning”) do not land well in the UK market in 2026. Reference something specific: the company's data maturity, a domain you understand, or a technical problem in their stack that genuinely interests you.

Round 2: Statistics and Probability (3 questions)

3. Explain a p-value to a non-technical stakeholder.

The trap is reciting the textbook definition. UK interviewers want to hear you translate it: it is a measure of how surprising your result would be if there were actually no effect, not a measure of how important the effect is.

4. How would you design an A/B test for a new feature, and what could go wrong?

Cover sample size and power, randomisation, a clear primary metric decided in advance, and the usual failure modes: novelty effects, peeking at results early, and metric definitions that shift after the test starts.

5. What is the difference between Type I and Type II error, and which matters more here?

Give a concrete example tied to a business cost: a fraud model's false positive blocks a genuine customer, a false negative lets fraud through. The “which matters more” half of the answer is the part that actually gets evaluated.

Round 3: Machine Learning and Coding (3 questions)

6. How do you know if a model is overfitting, and what would you do about it?

Cover the diagnostic (train versus validation performance diverging) and at least two remedies: more data, regularisation, simpler model, or better cross-validation, rather than jumping straight to “add more data” as a default answer.

7. Write SQL to calculate a 7-day rolling average of daily active users.

Tests window functions directly. Talk through partitioning and ordering out loud, and clarify how you would handle missing days before writing the query.

8. Walk me through how you would clean a messy dataset before modelling.

The interviewer wants a structured process: understand missingness patterns before dropping or imputing, check for leakage between train and test sets, and validate that a cleaning decision has not silently changed what the target variable means.

Round 4: The Case Study or Business Round (2 questions)

9. Customer churn increased 15% this quarter. How would you investigate, and would you build a model?

Strong answers resist jumping straight to modelling. Structure the investigation first (is it a definition change, a data issue, a genuine shift, which segment), and explain when a model is actually the right tool versus when the answer is a simpler analysis.

10. How would you decide whether a model is ready to ship?

Beyond accuracy metrics, UK interviewers listen for business framing: what does a wrong prediction cost, is the model fair across customer segments, and how will you monitor it once it is live. This question increasingly includes a follow-up on monitoring for drift, which sits close to MLOps territory.

Round 5: Behavioural and Stakeholder Fit (2 questions)

11. Tell me about a time a model you built was wrong in production.

The honest answer matters more than a perfect one. UK interviewers are listening for ownership, how you caught it, and what changed in your process afterward, not for a story where nothing ever went wrong.

12. Tell me about a time you had to say a model was not ready, against pressure to ship it.

This tests whether you will hold a technical line under commercial pressure. A credible answer explains the trade-off you made visible to stakeholders, not just that you refused.

A Quick Pre-Interview Checklist

refresh hypothesis testing, p-values, and A/B test design, out loud, not just on paper
prepare two portfolio projects you can walk through in five minutes, including what went wrong
practise SQL window functions until they are fluent, not just familiar
have a clear, honest answer ready for “tell me about a model that failed”
research whether the team's models run in production already, and what their monitoring looks like

The Honest Summary

UK Data Scientist interviews in 2026 test whether you can reason under statistical uncertainty, write working SQL, judge a model honestly rather than defensively, and connect a result to a business decision. Preparing for those four things covers the large majority of what actually gets asked.

Frequently asked questions

How many rounds are typical for a UK Data Scientist interview?

Four to six stages across two to four weeks: a recruiter screen, a statistics and machine learning technical round, a coding or take-home assessment, a case study or business round, and a behavioural round.

How much SQL is tested in a Data Scientist interview?

Enough to be genuinely fluent: joins, aggregations, and window functions in particular, since rolling metrics and cohort-style questions come up often.

Are machine learning system design questions common?

Increasingly yes at senior level, covering model monitoring, retraining triggers, and how a model fits into a wider production system, alongside the core statistics and modelling questions.

Data Scientist Interview Questions UK (2026, with Honest Answers)