Find me on
What’s the Impact? Experimentation and Causal Inference Get the Playbook
Free resources

Data science Python rounds: only 1 of my last 10 was LeetCode

I went back through my last 10 Python coding rounds for product and marketing data science roles, and only one was LeetCode-style

Data science Python rounds: only 1 of my last 10 was LeetCode. Here is what the other 9 looked like.
Bar chart of my last 10 Python rounds: recreate a stats concept 3, AI-assisted 3, pandas and data manipulation 2, interpret code 1, base Python 1.
Python round type 1, recreate a stats concept, came up in 3 of my 10 rounds, with pandas and numpy allowed.
Python round type 2, AI-assisted and open book, came up in 3 of my 10 rounds and was closer to a diagnostic case study than a coding puzzle.
Python round type 3, pandas and data manipulation, came up in 2 of my 10 rounds and combined data transformations with modeling.
Python round type 4, interpreting code handed to you, came up in 1 of my 10 rounds.
Python round type 5, base Python in a LeetCode style, came up in 1 of my 10 rounds, and syntax mistakes seemed heavily discounted.

I went back through my last 10 Python coding rounds for product and marketing data science roles, nearly all at Big Tech companies. Unlike SQL rounds, which tend to be fairly standard, these came in 5 distinct formats:

  • Recreate a statistical concept as a function (3/10). Pandas, numpy and other standard packages were allowed.
  • AI-assisted, open book (3/10). These were often investigative, closer to a diagnostic case study than a coding puzzle.
  • Pandas and data manipulation (2/10). Data transformations combined with modeling, either coding up a linear regression or talking through a modeling strategy for the data.
  • Interpret code handed to you (1/10). This was also a statistical concept in a function, but I had to interpret the code rather than write it.
  • Base Python, LeetCode-style (1/10). Syntax seemed heavily discounted here: I made several syntax mistakes but was clear about the logic and passed.

The obvious caveat is that this is a sample of 10 from one person, so the mix will vary with company, team and role. That said, AI-assisted challenges seem to be on the way in and increasingly likely to be a part of your interview process.

If you want to try one, there is a full AI-assisted coding challenge with the notebook on this site.

← All free resources

Every round in the loop, written up

The Playbook covers each of these formats in detail, along with case studies, statistics, applied machine learning, project deep dives and the behavioral rounds, drawn from the 80+ rounds I went through in Spring 2026.

See what the Playbook covers Other free resources