The Data Science Behavioral Interview Guide
Tell me about a time when...
From an employer’s perspective, hiring a data scientist is rarely about pure technical ability. To be effective, a new hire must communicate, collaborate, plan, prioritize, empathize, enable, and more! In other words, a successful data scientist must excel in areas that have absolutely nothing to do with a conventional data science skillset.
For this reason, you will almost certainly go through a behavioral screen as part of a data science interview track. While the questions may vary, the themes are consistent. Let’s get you ready.
Competencies
The most clichéd, terrible, no good behavioral interview question is, of course, “tell me about your strengths and weaknesses”. Hopefully you don’t get asked that - it’s incredibly unimaginative. Although, as an excellent candidate, you will of course have prepared for it.
Here is what savvier employers may look to measure:
Collaboration: Can you function in a team setting? Navigate difficult stakeholders? Know when to delegate and when to lead? It’s perfectly OK to be an introvert, or to prefer working alone, but employers need to know you can partner effectively with others.
Feedback: Can you give feedback? Receive feedback? Are you aware of the feedback others have about you? Can you deliver negative feedback with thoughtfulness and consideration? When you’ve received constructive feedback yourself, were you able to internalize and act upon it?
Impact + Influence: Are you able to connect your projects to business outcomes? Do you make sure your work doesn’t begin and end with you? Can you identify the right people in the company to put your recommendations into action? When have you convinced a skeptic?
Mentorship + Empathy: Especially important for senior roles. How have you helped others in their career? How have others helped you? Even in a competitive environment, do you celebrate and enable those around you? What will you bring to the team beyond just your own output?
Grit + Motivation: When the going gets tough, our data scientist should not get going (to another job). Data science projects have high uncertainty and an even higher failure rate, especially compared to disciplines like software engineering. Persistence is key to success, perhaps even more so than pure technical ability.
Self-Awareness: You’ve just finished an onsite interview panel. It’s the end of the day, and the hiring manager asks you how things went. It’s a trap! Don’t automatically regurgitate “It went great!”. The manager is looking to understand if your own feedback matches what the other interviewers are currently writing about you. If a particular session didn’t go perfectly, are you self-aware enough to realize it?
Planning + Prioritization: Do you sit down at your desk each day and work on whatever comes to mind, or are you a bit more intentional? When inundated with requests, can you focus on what’s most important? If you need help from others for a project to be successful, can you identify that gap and partner ahead of time? Do you think long-term, eliminating technical debt by ensuring your deliverables are well-documented and maintainable?
Values: Personal values are important to be a good human, of course, but in this context an interviewer is checking if you’ll demonstrate those of the company. Your author once had an entire behavioral interview consisting of “Tell me about a time where you exhibited <insert tech co core principle>”. Reading up on a firm’s mission statement and values should be part of any interview prep, behavioral or technical.
Scenarios
Prior to the interview, have at least 4-5 examples ready that can be modified on the fly to fit a given behavioral question. These experiences become easier to recall as you progress in your career, but even as a new graduate, try to come up with scenarios based on internships, group projects, research work etc.
Scenarios should have 4 aspects:
Context: Where are we? In school? A past job? If so, which one? Set the stage.
Problem: Perhaps you were dealing with a tough stakeholder, or were in danger of missing a deadline, or maybe you’d identified a bug in a coworker’s code. What was the issue?
Action: What did you do? Perhaps you sat down with a dissatisfied client to talk through their concerns. Maybe you had a difficult conversation to deliver necessary feedback, or coached someone through a hard time. Whatever the scenario, you need to have done something.
Result: Don’t leave us hanging! Make sure your example has a resolution. It doesn’t need to be a perfect outcome (in fact, messier examples are more relatable and effective), but we do need to have an ending.
Delivery
You’re in the interview. You hear the fateful words, “Tell me about a time when…”
Here are some tips:
Regardless of if a scenario immediately jumps into your head, pause to collect your thoughts. You never want to sound rehearsed, and it always helps to take a second to prep.
As you begin your answer, look for context clues. Is the interviewer following what you’re saying? Are they nodding? Frowning? Impassive? If the interviewer is confused, you may need to retrace your steps and give additional context.
Avoid technical jargon, acronyms, unnecessary domain knowledge. Your interviewer may not be a data scientist - they could be a stakeholder from Marketing, Sales etc. Ensure the person does not need to be an industry expert to keep up.
Utilize a scenario once, twice at most in a given session. Mix up your responses to show a diversity of experiences.
Be considerate of your scenario’s characters. For instance, when relaying an anecdote of a difficult stakeholder collaboration, don’t be too harsh on the individual in question. Part of a behavioral interview is about measuring emotional intelligence, which means you should always display empathy and compassion in the retelling, regardless of how you felt in the moment.
Hacking Data Science provides practical career advice for data scientists, aspiring and practicing. You’ve just read a free weekly article. Paid subscribers get access to two more paid-only posts each week, along with our full archive.