Preparing for a Data Scientist Interview in a Structured Way
Competitive success in many industries is becoming increasingly dependent on the ability of businesses to exploit data science. With the growing importance of data science there has been a corresponding expansion in both the hard and soft skills expected of Data Scientists.
There are numerous websites which provide long unstructured lists of example interview questions. Rather than repeat these, we’re going to provide you with some structure to help you get prepared more systematically and more comprehensively.
Before we look at the skills interviewers are seeking, it’s useful to be aware of the types of questions interviewers often use.
- “Explain” questions. Most often used for probing technical skills (e.g., ‘When and why should one make use of ridge regression?’)
- “Relate your experience” questions. Past behaviour is a reliable predictor of future behaviour. Make sure you have two or three impressive past experiences ready to hand that you can use to answer any variety of this type of question (e.g., ‘Tell me about a time when you solved a business problem using data science / taught yourself a new data science technique and made use of it / overcame an interpersonal difficulty when working in a team, etc.)
- “Solve a problem” questions, which test your critical thinking skills (e.g., ‘How could you use data science to enhance the call centre experience of our customers? Practically, how would you go about it?’)
- “Creative or left-field” questions. Arguably the toughest type for many of us, these test your imagination, your curiosity and (frankly!) your ability to deal in-the-moment with the unexpected (e.g., “Which industry do you think is going to be most transformed by data science and machine learning?”, “How far away do you think we are from human-level intelligent machines?”)
So here’s the broad categories of skills you may be asked about. Think of these as layers of an onion, from a core of hard technical skills to the outer layers of probing your softer personal and interpersonal attributes. Speaking of onions, never forget that any interviewer who knows their onions will be as interested in how you go about answering their questions (how structured you are, how well you communicate, what energy levels you project, etc.) as in the actual answers you give. Here we go:
- Layer 1: Technical skills
- Core data science approaches (algorithms, techniques, etc.)
- Coding languages (Python, R, SQL, etc.)
- The data science pipeline (data preparation, feature extraction, training and testing, deployment, monitoring, etc.)
- Software tools and platforms (e.g., data cleansing, visualisation, model monitoring, etc.)
- Layer 2: Critical skills that amplify your core technical skills
- Ability and willingness to learn. Curiosity
- Business understanding
- Layer 3: Soft skills that enable you to be successful in an organisational context
- Communication, both written and oral
- Collaboration, relationship-building
- Drive, motivation, passion, resilience
- Discipline, systematism, organisation, time management
The balance of which skills are more likely to be covered will depend on the nature and seniority of the role. More junior roles are likely to be biased towards layers 1 and 2, more senior roles to layers 2 and 3.
All of the question types we reviewed can be used to test each of the skills described above. We’d recommend working through the list of skills one-by-one, looking for example questions online and inventing some of your own and then rehearse, rehearse, rehearse.
Lastly, never forget that an interview is bidirectional. Passively answering questions rarely trumps over doing that and actively demonstrating energy and enthusiasm (and even some personality!) through a lively personal introduction, asking questions, and telling compelling personal stories.