I've scouted smaller/less known schools with great CS programs, so it definitely can be done.
I know enough French to be dangerous so they sent me to the Polytechniques (Montreal, Paris, Swiss...) and got a great signal to noise ratio there. But it's a plane ride instead of an Uber...
> Better to just look for things like: grit, coachability/potential, focus, ability to reason about things across many orders of magnitude, ability keep a lot of aspects of a complex system visualized, and generally whether they can understand problems and come up with solutions. None of these involve playing "algorithm gotcha" or solving leetcode puzzles.
A degree from a solid school is a good predictor of that.
Algo questions are used because it's a common vocabulary, everyone who matters just knows this stuff and it's possible to solve a problem in 45 minutes (vs giving a 2hour+ take home assignment). And it's really good at spotting people who, no matter how long you give them, won't be able to write a single line of code.
If you want to hire 100 people then using "Got into Stanford and didn't flunk out" is a fairly cheap filter to approximate these things.
https://www.usnews.com/education/best-colleges/articles/how-...
Yet, when [the usual suspect] companies are hiring, they back a truck up to Stanford and just shovel new grads onto it. Or their automated filters (that desperately try to get the candidate flow from 6,000 to 6) eliminate anyone that doesn't have a narrowly defined "right background".
Better to just look for things like: grit, coachability/potential, focus, ability to reason about things across many orders of magnitude, ability keep a lot of aspects of a complex system visualized, and generally whether they can understand problems and come up with solutions. None of these involve playing "algorithm gotcha" or solving leetcode puzzles.
EDIT: Easier said than done, though. I don't have a practical solution to how to do this with a multi-thousand candidate flow for each job posting.