Data Scientist mock interview
Answer common data scientist interview questions out loud, from “Tell me about yourself” to model choices and experiment design, then get feedback on what you said.
- Free, no account
- Answer by voice or by typing
- About 10 minutes
What they assess
What Data Scientist interviews look for
Statistics and experimentation
Whether you understand the methods behind your work: hypothesis tests, experiment design, bias, and what a result does and doesn’t tell you.
Machine learning judgement
Choosing a sensible model for the problem, evaluating it properly, and knowing when a simpler approach is good enough.
Problem framing
Turning a vague business question into something you can measure or model, and being clear about your assumptions.
Communication
Explaining models, uncertainty and trade-offs to the people who will make decisions from them, without hiding behind jargon.
Getting to production
Working with engineers and product teams so a model is used, monitored and improved, rather than left in a notebook.
Questions
8 common Data Scientist interview questions, and how to answer them
- Opening
Tell me about yourself.
What they’re looking for: A short summary of the data scientist you are: the problems you work on, the methods you use most, and one project that had a real effect.
How to answer: Start with your current role and the kind of data and models you work with. Give one project and what changed because of it. Finish with why this team’s problems interest you.
- Behavioural
Tell me about a model you built that made it into production.
What they’re looking for: End-to-end ownership: framing the problem, choosing and evaluating the model, and what happened once people used it.
How to answer: Explain the business problem and how you framed it. Cover the data, the model you chose and why, and how you judged it was good enough. End with the result in production and what you monitored.
- Behavioural
Tell me about a time your analysis or model didn’t work as expected.
What they’re looking for: Honesty and learning: how you noticed, how you found the cause, and what you changed afterwards.
How to answer: Describe what you expected and what actually happened. Walk through how you worked out why (data leakage, drift, a wrong assumption). Finish with what you did next and what you now do differently.
- Behavioural
Tell me about a time you explained a model’s results to non-technical stakeholders.
What they’re looking for: That you can make uncertainty and trade-offs understandable, and help people reach a decision rather than just admire the model.
How to answer: Say who the audience was and what they had to decide. Explain how you framed the result, including its limits and how confident you were. End with the decision and what you’d present differently next time.
- Role-specific
How would you design an A/B test for a new feature?
What they’re looking for: A clear process: a hypothesis, a primary metric, randomisation, sample size, run time, and how you’d read the result.
How to answer: Start with the hypothesis and the one metric that decides success, plus guardrail metrics. Explain how you’d randomise, estimate the sample size and decide how long to run it. Finish with the pitfalls you’d watch for, like stopping early or testing too many metrics.
- Role-specific
How do you choose between a simple model and a more complex one?
What they’re looking for: Judgement about trade-offs: accuracy against interpretability, cost, upkeep and the needs of the people using it.
How to answer: Say that you start with a baseline and only add complexity when it clearly helps. Name what you weigh: performance on the right metric, explainability, latency and maintenance. Give an example where the simpler model won.
- Role-specific
Explain overfitting and how you prevent it.
What they’re looking for: A plain explanation and practical tools, not a textbook definition.
How to answer: Describe overfitting in a sentence: a model that learns noise in the training data and does worse on new data. Then say how you detect it (a held-out set, cross-validation) and how you prevent it (simpler models, regularisation, more data, better features).
- Motivation
Why do you want to work here?
What they’re looking for: That you know what problems the company solves with data and why they interest you.
How to answer: Name a specific product, dataset or problem the company works on. Connect it to your experience and the kind of work you want to do more of. Keep it about their problems, not just their tech stack.
More Data Scientist interview questions to practise
- Walk me through a project on your resume from start to finish.Opening
- Why data science, and why this kind of role now?Motivation
- Do you prefer research-style work or shipping models into products? Why?Motivation
- Tell me about a time you worked with a messy or unreliable dataset.Behavioural
- Tell me about a time you pushed back on a request because the data couldn’t answer it.Behavioural
- Tell me about a time you worked closely with engineers to deploy a model.Behavioural
- Tell me about a time you had to learn a new method quickly for a project.Behavioural
- Tell me about a time your recommendation wasn’t followed. What happened?Behavioural
- Tell me about a project where you had to balance speed against rigour.Behavioural
- How would you explain a p-value to a product manager?Role-specific
- What evaluation metric would you use for a fraud detection model, and why?Role-specific
- How would you handle a heavily imbalanced dataset?Role-specific
- What is data leakage, and how have you guarded against it?Role-specific
- How would you tell whether a model in production is getting worse?Role-specific
- An experiment shows a positive result, but a guardrail metric dropped. What do you recommend?Role-specific
- What questions do you have about how data science works here?Closing
All 24 questions on this page are in the practice above; each round picks five.
Mistakes to avoid
- Spending most of the answer on the model and almost none on the business problem or what changed.
- Quoting accuracy alone without saying why it was the right metric for the problem.
- Using jargon with a non-technical interviewer instead of explaining the idea plainly.
- Presenting a project as a success without mentioning its limits, assumptions or what you’d improve.
Questions
About Data Scientist interviews
How the practice works and what happens to your answers: about this mock interview.
What is a data scientist interview usually like?
It varies a lot by company. Common parts are a recruiter call, a technical screen (statistics, machine learning concepts, often SQL or Python), a case study or take-home task, and a behavioural round. This practice covers the spoken parts: behavioural questions and explaining your methods out loud.
How much statistics should I expect?
Expect at least the basics: hypothesis testing, confidence intervals, experiment design and common pitfalls. Product-focused roles tend to lean more on experimentation, and machine learning roles more on modelling and evaluation. The job description usually gives you clues.
How do I talk about course or personal projects?
Treat them like work projects: the question, your approach, the result and what you’d change. Be clear about where the project came from, and spend your time on the choices you made rather than the final score.