This year I reviewed more than 2,000 take-home assignments

This year I personally reviewed more than 2,000 take-home assignments for data analyst positions.

I’m a big fan of this approach as it helps us to choose better candidates.

AI changed how we design these tasks. We need to ensure that it can’t be solved by AI alone with a couple of simple prompts. But I don’t do this to filter out people who use AI. I want them to use it. The best candidates are the ones who use AI well, not the ones who avoid it.

An example of the tasks we give to Junior Analyst candidates is to recreate a simple dashboard in any BI platform. It helps us assess data literacy, accuracy and data visualization skills. I’m really happy when candidates use AI to do it. But only a small number of them manage to do it well, even with AI.

What else does it show us? Whether a person is hard-working. In a startup we need people who can go the extra mile to get the result, and a simple task like this makes that easy to see.

A good assignment should be relevant to the actual job, and should clearly communicate the expected effort. We then discuss the candidate’s decisions, trade-offs, and reasoning during the interview.