Tips For A Data Science Interview

Learn 6 Tips For A Data Science Interview

By: rkspark

Data science is such a huge area right now and a job in the arena is an excellent prospect
In fact, according to Harnham, data science has significant prospects for the future. So, how do you ace an interview?
1) Have your stories ready: They should be both prepared and rehearsed. You’re likely going to get hit with questions that try and learn your character and attitude, so you need several instances from your past you can relate to in order to demonstrate traits that are desirable. For example, Amazon provides a list called “Leadership Principles” which they like to interview against. Find this or something like it, and then spend a few hours reflecting on possible answers based on good things you’ve done in order to come up with concise answers to these kinds of questions. The trick is being able to deliver a short anecdote that gives just enough information to create a narrative painting you as a person that does good things. You might get asked clarifying questions, but don’t ramble or over-provide information. Being boring works against you.

2) Minimize the surprise: One of the most difficult aspects about interviewing is needing to think quickly so you can get to answers while you’re under a microscope of attention and on a clock. Even questions you might consider normally easy are hard under such conditions. Reduce the surprise as much as you can by learning what your interview is likely to involve so you’re prepared for everything you know is coming. Glassdoor has the interview reviews and results of quite a few folks that have already been down this road, so never take any interview until you’ve done your homework and learned what’s out there. Always talk to your recruiter about what you can expect, as well as any advice they might have for getting ready.

3) Practice problem-solving: Do it out loud, but also do it by writing solutions down on paper. If you are a Data Scientist, then you can expect a lot of SQL questions. Use ProgrammerInterview or JitBits for sample questions so you can actually do work on them. If you make any mistakes, jot down what you specifically missed so that you know what topics to spend more time on.

4) Establish your framework for problem-solving and then practice it: In the many technical interviews I have done, I’ve learned that a similar or identical framework is applicable to most all questions, and that clarity of thought provides me momentum. I could dedicate an entire article to laying out the particular framework that I personally use, but you can develop one of your own just by working a lot of the available problems.

5) Know the fundamentals: The combinatorics pages at are a great way to get a refresher on the core principles in addition to providing examples you can use for self-testing. It’s a good idea to be prepared for prob and stat questions, so don’t waste any time trying to recall Bayes’ Theorem when you can just be ready for it in advance.

6) Know your common technical concepts: If you are in Data Science, then you have to comprehend bias/variance. You have to know the many ways of detecting and/or handling an overfit model. You’ll also have to know various strategies for handling classification problems when any classes have a high imbalance. You also need to be aware of the advantages and disadvantages of the different model structures, and you need to have a firm understanding of the basics of just how algorithms work. Do you know what’s random about random forests? Can you say what’s meant by something like gradient boosting? Treat your technical interview just like it’s a comprehensive, oral, final exam. Identify applicable major concepts and learn their core principles.

Back to Top