Interview scorecard
Data Scientist
Score each skill from 1 to 5. A 1 is below the bar for this role, a 3 meets it, and a 5 is above it. The examples show what each sounds like. Use 2 and 4 for answers that fall in between.
Statistics
Applies statistical reasoning to draw sound conclusions from messy data and explains uncertainty plainly.
Questions
- Tell me about a time statistics changed a conclusion people were ready to accept.
- Explain a confidence interval to a manager who has never studied statistics.
- Tell me about a time your data was biased or incomplete. What did you do?
1: Below the bar
Quotes methods or p-values without explaining them. Treats a single number as certain and ignores bias in the data.
3: Meets the bar
Explains core ideas correctly in plain words. Notices common problems like small samples or bias, sometimes only when asked.
5: Above the bar
Explains uncertainty simply with a real example. Spots data problems unprompted and says how they would change the conclusion.
Modeling
Builds predictive models suited to the problem and checks that they work on new data.
Questions
- Tell me about a model you built. Why did you choose that approach?
- Tell me about a time a model performed worse than expected. What did you do?
- How did you decide your model was good enough to use?
1: Below the bar
Picks a method because it is familiar or popular. Cannot say how they checked it works on new data.
3: Meets the bar
Explains why the approach fits the problem and tests on held-out data. Handles problems with some guidance.
5: Above the bar
Compares simple and complex options, explains the choice, and finds why a model failed. Links the quality measure to its real use.
Experimentation
Designs and interprets tests that show whether a change caused a measurable difference in results.
Questions
- Tell me about an experiment you designed. How did you decide what to measure?
- Tell me about a test where the result was unclear or surprising. What did you do?
- Describe a time you could not run a proper test. How did you still reach an answer?
1: Below the bar
Compares before and after without a control group. Takes any difference as proof and cannot say what else could explain it.
3: Meets the bar
Sets up a control group and a clear success measure. Explains results correctly but may miss sample size or side effects.
5: Above the bar
Defines the measure and test size upfront, checks for flaws, and explains what to do when a test is impossible.
Programming
Writes clear, working code that others can read, rerun and trust to analyze data.
Questions
- Tell me about a piece of analysis code you wrote that others had to reuse.
- Describe a time your code gave a wrong result. How did you find the cause?
- How do you check that your data cleaning code does what you intended?
1: Below the bar
Code works only on their own machine. Can't describe how they test it or find bugs.
3: Meets the bar
Writes readable code others can rerun. Checks results by hand and can describe how they tracked down a past bug.
5: Above the bar
Writes tested, reusable code with clear structure. Checks data at each step and explains how they caught and prevented bugs.
Business understanding
Connects data work to company goals, customers and money, and shapes analysis around real decisions.
Questions
- Tell me about a time you changed an analysis after learning what the business needed.
- Describe a project where you chose what to work on based on its value. How?
- Tell me about explaining a result to someone who had to make a decision.
1: Below the bar
Talks only about methods and data. Can't say who uses the results or what decision they support.
3: Meets the bar
Names the decision and the stakeholder behind a project. Links results to a business goal but gives vague impact.
5: Above the bar
Starts from the decision, weighs value against effort, and shows measured business impact. Adjusts the message to the audience.
Overall recommendation
Decide on your own, before you compare notes with the rest of the panel.