Primer on Machine Learning Interviews for Software Engineers

Indeed the role of a software engineer is evolving with time as GPTs are getting integrated into codebases and amplifying engineers workspace. So far it looks like an incremental update and not a paradigm shift in the workforce, where AGI isn’t doing our daily jobs and is more of a productivity multiplier.

It’s unclear to me¹ yet how things are going to look in 10 years, but in the meantime I think the projection of how interviews for machine learning engineers (MLE) is rather static. I would assume that companies will still look for great engineers with strong fundamentals, and the amount of expectation for code will increase. Software engineers with strong system engineering backgrounds will probably be able to do the same job that an entire team does now!

The best way to think about MLE interviews is a software engineering interview on the lookout for people with strong backgrounds in machine learning. There is no one-size-fits-all definition of MLE interviews, as depending on the size of companies and the nature of their product, they tend to have different priorities.

From my experience, especially at the most competitive places, they’re basically looking for the best coders/system engineers that also happen to have a strong background in machine learning. Therefore interview preparation should be an optimization for software and machine learning.

Typically, expect to be interviewed around:

1. Software Engineering²: data structures and algorithms.
2. Mathematics: statistics and probability.
3. Machine Learning: classical algorithms and techniques.
4. Neural Networks.
5. Systems Engineering.

Each company does it differently, look for different signals, etc so the best thing you can do is prepare³ for the toughest ones.

Notes