3 ms·
I think it could, in theory. I've had the idea of approaching it more like a sorting problem. Scrape social media of the candidates and the company, use NLP to
by SuoDuanDao 4y ago
I think it could, in theory. I've had the idea of approaching it more like a sorting problem. Scrape social media of the candidates and the company, use NLP to provide a 'cultural fit' metric that circumvents the interview process...
- fallingknife 4y agoProblem with culture fit is that no one is willing to be completely honest about what they want, or what their culture is. No candidate is going to ever say "I'm a mediocre engineer really just looking for something easy where I can coast. Below market comp is OK." And no company is ever going to say "Kind of bureaucratic for a company this size, so want somebody who has a high BS tolerance, and will put up with that in exchange for good pay and reasonably low productivity standards." Also god help you if your culture fit algorithm has correlations with race / gender / nationality. And there's pretty much a 100% chance it will.
- ipaddr 4y agoMediocre means average. Most developers/people are mediocre or average. Perhaps you mean below average.
- bigbillheck 4y agoSounds like an even worse approach than Triplebyte's.