Hacker News Reader

Comment by dariusj18 | Hacker News Reader

My experience is it's impossible to find people who's experience actually matches their resume.

I've heard this so many times....But, i believe that there is always the possibility that someone who is not a perfect match can evolve and learn to become the perfect match. I've also seen the case - though rare - where the "perfect candidate" fails to evolve to learn new things, or fails to adapt to new conditions because they're too intsrested in only a specific scenario. I've been on both the candidate side as well as (many times) the hiring manager side. I think too many HR departments and hiring mangers lack the mechanism to gauge candidates, so they fall back to metrics that might be too generalized. Granted, a non-standard approach to finding that candidate who can evolve is not super-scalabel - i get it - but its still my preferred direction. In all cases, when i have hired using my approach, i have always had great candidates; my approach has yet to fail me. Meanwhile, my peers who hire using too rigid a set of metrics...well, let's say that they've gotten good candidates only about half the time.

That is the case when hiring entry level and junior, but a senior level should be able to hit the ground running on at least one part of the job requirement.

Agreed, that is true. A senior does not need to meet all requirements but at least 1 or 2, yeah.

We try to hire seasoned linux admins for our ops team. This has become difficult.

I start with this technical interview question so I can calibrate quickly:

"If I run the following two commands:

mkdir a b c;mv *

What happens?"

One person so far has asked if globbing was active. Like half of people get the answer right if they assume it is.

This question is deemed as "not fair", though, because it assumes a linux admin would know about shell globbing. So we're trying to find a modern equivalent that is somehow relevant, even though we're literally hiring people to know some shell.

You didn't ask for advice, but I have some Linux experience and it does feel a little like a trick question since you would normally never do that. Probably I'd give 100% credit for answering the question as asked, and 90% credit if they can explain to you what happened after running it in a terminal or if you tell them the end result.

If you need good questions, maybe ask your current admins the five most common operations or pain points they have every day and then base your questions on that.

Yeesh this feels more like a trivia question and less like a real problem.

You should be focusing on a desired objective in the linux environment and then allow them to outline the steps to get to that objective.

If half the people get the answer right, what's the problem?

If the question is irrelevant or about unimportant trivia then it's a bad signal to use to reject people. An extreme example would be rejecting any candidate who fails to call a coin flip correctly.

Same. The tech exercise we give to candidates is to write two very simple CRUD end points in Django/DRF, and to add date filtering to one of them. There are various ways to implement this and I’d accept a lot of variation in the final solution, but probably 80% of them didn’t work correctly. And that’s from people that mostly say they “know” Django on their CV.

Do your job posts match the jobs?

This only comes into play once you've started the interview process, unless you're saying you're so jaded that you don't even bother contacting candidates with good resumes.

Reply on news.ycombinator.com