3 ms·
I will say as someone who's been involved in the hiring side, the 100s of applicants number is true but also misleading. Especially if you take public applicati
by Macha 3y ago
I will say as someone who's been involved in the hiring side, the 100s of applicants number is true but also misleading. Especially if you take public applications on linked in or someone. You'll get a lot of people who are unqualified, and I don't mean stuff like "3 years of experience and the req said 5" or "Only worked in scala but we're hiring for python devs", but rather "This person is a cashier and has not indicated anywhere they would have gotten development experience", or "This person is a fresh graduate applying for a team lead role".
Then if there are still too many to interview after that first filter there's the easy ways to cut down numbers further like "This person has no visa and has not started any visa process to work in our jurisdiction" or "this person's entire career is in a single tech stack that is not our tech stack while there's other candidates with our tech stack or a track record of adopting new ones". Could these people work out? Possibly. But it's an overhead that when there's extra candidates, they need something to stick out to make it onto the shortlist. If there's others that have all the same pros and an already valid visa/citizenship or the right tech stack, they'll get on the shortlist first.
After those two processes, the number of applicants per role are not that crazy.
- TheCaptain4815 3y agoThis would be such a fantastic use of LLMs. Each step you listed to curtail your list could be done fairly easily using a specific model (prob classification oriented). I wonder if there's a market for something like this.
- Macha 3y agoThere's already automated systems that do this. We do it manually just because we don't want to risk a system automatically discarding a candidate we might actually like.
- TheCaptain4815 3y agoThat's kind of my point, I wonder if LLMs (especially something new like GPT4) could surpass those 'old school' automation methods for this given their 'understanding of context'.
- DonHopkins 3y agoThen applicants will start jailbreaking the LLMs in their cover letters to get to the front of the line. Which will be great if you're trying to hire prompt engineers.
- DonHopkins 3y agoInhumane Resources
- Uptrenda 3y agoI am always curious what kind of candidates the competition are. You have a unique insight as someone who does hiring.
- b20000 3y agothat is why you have to first verify whether they have a master's in CS, if your requirement asks for that. you call the university and find out. if they don't, you can discard the candidate. then, you can filter on the visa requirement etc, and only after doing all these obvious things you go on to your requirements. this also means you don't need to do leetcode interviews. anyone from a decent university with an actual degree will be able to code, or quickly learn how to. you said it yourself - too many candidates, so no time to do leetcode BS.