Co-Occurrence & Recommendations

Affirm sits in the half-second between a shopper and a checkout button on thousands of merchant sites. Someone splits a Peloton into four payments. Or a Casper mattress. Or a Wayfair order. Each time, Affirm learns one small thing about that person: the set of stores they shop. Stack that up across millions of shoppers and a graph appears, and the graph says which merchants get bought alongside which. Turning it into a recommendation for the next shopper is a real problem Affirm's engineers work on. It is also, nearly word for word, one of the two coding questions the company reaches for most on a phone screen.

That makes it worth more than the usual practice reps. It recurs. Candidates have run into this question again and again, and only once did it show up in Affirm's own merchant language. The other times it wore a costume: fruit in a shopping basket, letters that share a word, a puzzle handed over as Max Company Count. Same problem every time. Learn the shape once and the costume stops fooling you.