Hacker News Reader

Comment by ilrwbwrkhv | Hacker News Reader

I wish interviews were like this. Show them something you built and you get hired instead of the nonsense of leetcode.

In fairness it isn’t a great filter. As someone that had a similar story - I got a very well paying job out of high school based in part on some home grown apps 20ish years ago - the field of people doing small scale stuff was small and any basic CRUD app, let alone embedded hardware hacking made you instantly look like a wizard - since getting started then was much harder - any output was proof of a certain level of skill in acquiring knowledge that was above average even if the overall scope wasn’t huge. In this day of Arduinos/RPi and git pull boilerplate, a small app or hardware project doesn’t prove much of anything. The scale of side project to really differentiate is beyond what most people reasonably have time for if they wish to have a life. I think compensated take home assignments still have some merit though.

FWIW, that's how I structure my interviews typically, so they still exist in some places.

For anyone reading along I highly encourage that; there's way more signal to noise hearing about some project they worked on and why they were proud enough to put it on their resume than 'did they remember the algorithmic call/response'. You can't beat an almost post mortem discussion where you let the interviewee lead on what they think went well and what they'd have done differently with hindsight.

Do you find it works even if they can't actually show you any code, because it's proprietary belong to other employers?

Yes. If people talk enough they can’t help but share all the relevant details if you know enough about creating medium or large sized systems. Also I ask questions for stories about troubleshooting bugs, and when to log, some story when logs And I would rather hire someone that iterates well over the weeks or months time frame than can solve something quickly.

For newbies, it’s harder.

100% this. All you really have to do is get people talking. Have them tell you a story and ask for details. If they have trouble when you get into details then they are making it up. If not you should have a fun conversation.

As a recent interviewee I can say that I'd be asked about code I was proud of. The last place I worked for was well-known enough or could be looked up easily and was small, so I explained what our main solution did, a particular issue we faced and how I planned and implemented a fix. Led to some good conversations with pseudocode, and didn't break any NDAs.

Shockingly low percentage of people have built something they are proud to talk about. All them were good hires tho, as long as the explanations are detailed enough that it was work they actually did or adopted well enough.

Even the stuff we are proud of isn't possible to talk about due to NDAs.

If you put a little work in, you can talk about nearly every project. Even stuff like classified satellites have been discussed; you just act respectful to the interviewee's boundaries (which you should be doing anyway).

I try to drag this out of candidates, but it's often very hard... Surprisingly few people actually code outside their jobs/education... Job code is obviously a no-go, and academics are usually pre-skeletoned work, So when I ask them to show me their favorite project, they often draw a blank...

I was lucky and I did a side project my last year of college abusing Hadoop to make a web scraper & analysis tool on one. Probably got me my first 2 jobs because I LOVED to talk about that project, and it let me break out of my fearful and introverted shell to show a wider range of skills

I got hired on at a Fortune 500 company in the early 2000s. One of the reasons I've stuck around (besides the great pay and benefits) is that I've read all the horror stories of leetcode interviews. I have zero interest in participating in that nonsense. One benefit of this is, when I am interviewing people to fill development position, I never do any of that leetcode nonsense.

You should still check out to see if your pay is competitive for your role, level, etc.

Even an awesome environment might be worth moving away from if your pay goes up 50-100%. Especially if you're not super frugal or have expenses due to other reasons.

The problem is I don’t want to move (great house, low mortgage, great area, etc.). There really aren’t any other major companies within 300 miles and all the smaller companies in the area don’t pay anywhere close to what my company pays.

But is it worth doing leetcode? I find I can’t bring myself to do it even if it would guarantee me a FAANG position and probably triple my salary overnight.

Overengineering starts with the interview.

It's probably not worth it on its own.

But if that tripled salary enables you to do things you love you can't do now and it enables them both today and long term... then it might be worth it :-)

Well you can interview while you have your job and no one needs to be any wiser. You can be sure that at least annually the company is looking to stay lean and mean and right size you if it fits their economic goals for the company.

Requiring coding tests etc just wasn't filtering more than random chance when I was hiring. I ended up pulling in people who had the most preposterous CVs to see what they were actually like.

One guy was a ninja. A black belt in ninjitsu, apparently. I asked him about his ninja skills (he looked like Harry Knowles). He said he could make himself invisible. I was like O_O. I asked him to demonstrate. He said he couldn't do it right now as we already knew he was there. I told him to leave the meeting room and "come back in invisible." He said it didn't work like that. I asked how it works. He said "Let's say I'm at a party.. I can enter the room and walk completely through the crowded room and not a single person will notice me." I think I know what was happening....

My story upthread is pretty much the same, except that I demonstrated a record-keeping application written in Turbo Pascal :-)

My first program (that wasn't just printing 'you suck' over and over in basic on the C64) was in Turbo Pascal. It used showed how light defracted in different materials using Snell's Law.

It was great.

For my very first interview I was told “write something in php” so I put together a few scripts to parse templated txt files into html and made a little baby version of a cms. Wasn’t exactly scalable or ground breaking but I thought it was cool. But the engineer that interviewed me was like so this imports text files and it took you all afternoon to build it? …didn’t get that job.

So the next one was pretty similar “hey use this language called Sikuli to build something.” Now, I’d never heard of Sikuli but it was basically python so I built a little app that, based on the screen, could offer up context sensitive hotkey chains…kind of like a baby version of Alfred that literally only worked with Microsoft office. But the guy thought it was cool and hired me anyway :)

Went from working in a gas station over night to working in my kitchen or the coffee shop. I thought it was sooooo cool

One problem is that the big companies (that, for the most part, pay the big bucks and people care about and other small companies look to) are market movers. If they are known to be doing some alternative to leetcode then leetcode-type websites and study programs will pop right up to optimize for whatever new metric they have. If they start to prioritize things you've built over ability to perform, you can bet your bottom dollar that within a week there will be a million guides on crafting the optimal cookiecutter project to get in the door at Facebook available on the web. And the signal will lose its value. A signal has to be costly to acquire and algorithmic prowess, although made easier through things like leetcode, is costly for everyone to acquire which is why it holds its value.

The few interviews I had were exactly like that. Not sure if it's a regional thing or if I have just been lucky.

But telling to story how young horny me built porn sites and learned SEO the hard way was actually a selling point so far.

I'm thinking of changing jobs and the need for me to do leetcode and answer trivia gotcha questions vexes me. More than 10 years of professional experience and like 5-6 years of hobbyist experience as a web master for 00's websites, and managing ecommerce sites with >$1M's in revenue doesn't count for anything.

It is if you’ve built something.

Most people don’t have much to show.

Reply on news.ycombinator.com