Restricted category To view older posts, you must first submit an approved post in that category

Data Science Interview Tips

Hello Candor community. Data Science interviewer at a big tech company here!

Providing some tips for people who may be looking to start preparing for DS interviews as hiring ramps up in 2021.

DS interviews test 3 primary skill types: programming skills, theoretical skills, and business/product acumen.

Programming skills

As a DS, because you aren’t writing production level code, this isn’t a deciding factor in your interviews. However, you should still review your algorithm questions. Spend some time practicing Leetcode questions to show your interviewer you have the necessary foundational skills.

Theoretical skills

There are two types of theory questions. The interviewer might ask “rapid fire” questions where they check your knowledge of foundational concepts (breadth), or it might be a deep dive into a concept—main question will follow ups (depth).

Rapid fire examples

  • What is the difference between exploration and exploit?
  • What is the loss function of SVM?
  • Can random forest handle imbalanced data?
  • If we reverse x and y in single variable regression, how will Rsquare change?
  • Write out variance, MSE and bias, and tell me the difference between the three

Deep dive

  • Whiteboard interview where you are asked to explain and calculate concepts
  • Think of it as a applying the rapid fire questions to a real example

Business/product acumen (paired with technical skills)

These questions will pair your technical skills and test your business and product intuition—what the real job will be. It doesn’t matter how great your technical skills are if you can’t apply your knowledge to real world data. Here are some example questions (different levels of technical focus):

  • Please write your own function to implement a KNN algorithm
  • Please write a function to implement Ridge Regression
  • Facebook is studying the relationship between the number of friends and user engagement. Write SQL code to count the number of times people who are friends in the data set engage with each other.
  • Uber discovered that one of their algorithms has a negative effect after being put into production compared to their out-of-sample test. What are the potential causes and how would you troubleshoot as a DS?

A common way companies test this last category is with a take home assignment. The assignments test your problem solving and modeling skills, in addition to your business intuition. A company can tell right off the bat if you’ll be a good fit (skill/mindset wise) so make sure you clearly think through how you approach the take home assignment.

Hopefully these insights and sample questions are helpful and best of luck to everyone looking into DS jobs/internships!


Thanks for sharing your tips and sample questions!

This is really useful. I shared some of the “rapid fire” questions you described here: Google Data Science Interview Questions

Super helpful to see posts from interviewers on here. Really no way to get better/more direct advice! Thank you!