Data Scientist Coding & Algorithms Interview Questions

Coding in a Data Science loop is not the Software Engineer bar, and these 331 questions show where the line actually falls. You are asked to implement the method rather than tune a data structure: write univariate linear regression by ordinary least squares from scratch, simulate coin flips until the empirical distribution tells you whether the coin is fair, work out the expected number of good pairs in a randomly weighted graph, multiply two numbers without the multiplication operator. Record processing appears in shapes a statistician meets in the job, such as deciding whether two streaming trade feeds are equivalent when ordering only has to hold per company. A heavier strand from quant and infrastructure teams asks for tiled matrix multiplication in CUDA and an explanation of the memory hierarchy that makes tiling pay. Amazon, Google, Pinterest, Two Sigma, Uber and Capital One supply most of the set. Only 36 are rated hard against 253 at medium, so the ceiling sits below a backend screen, yet 43 arrive as take-home projects where tested, readable code counts for more than speed. 276 open in a console, and 84% are readable without a premium account.

331 Questions 80 Companies09.28.2026
Showing 20 results

Explore more Data Scientist Coding & Algorithms interview questions

Real questions from candidate reports, grouped by company, topic and role.

By company
Other categories for Data Scientist
Coding & Algorithms questions for other roles
Browse all