Why we hire for judgement, not syntax
Our interview process has no whiteboard puzzles. It has a paid trial on real work, and here is the reasoning behind that choice.
Algorithmic interviews measure preparation for algorithmic interviews. They correlate with having had time and money to practise, which in this market selects for background rather than ability.
What we actually need to know is whether someone makes good decisions under ambiguity: whether they ask the right question before writing code, whether they know when a simple solution is sufficient, and whether they will say so when a plan is wrong.
So the process is a conversation, a practical exercise drawn from work we have genuinely done, and for senior roles a short paid trial engagement. The trial is paid because asking for unpaid work filters for people who can afford to give it away, which is exactly the filter we are trying to remove.
It is slower than a puzzle round. It has also been considerably more accurate.