Hacker News Reader

> 1. AI Spam. I categorized the inbound we got the other day from a job post. Out of 172 daily applicants, we got 22 that looked reasonably like a person, and 150 that were primarily AI generated messages. Which are pretty easy to spot because they're 500 words of tech jargon and rehashing the job description.

This is what it takes to get through the filters/recruiter search. In majority of the organizations the first line of defense is the recruiter who has a limited knowledge of the job responsibilities, technologies involved, etc. They employ various search techniques in their ATS (lately enhanced by the various LLM tools) and whatever resumes come up in the keyword search will end up in the pile presented to the hiring manager. This is especially true now when volume of resumes is in hundreds, you can't just go through all of them manually. So the only way to get through the first stage is to create resumes to fit the LLM/keyword search...

EDIT: I've been seeing more and more of these disclaimers when I submit an application: "We use Machine Learning for an initial comparison of resumes against the education, experience, and skills requirements of the job description." Check here to out out of this analysis.

So candidates tweak resumes to match the requirements (likely) with the same LLM that will be checking the resume match to the same requirements. Why is this surprising?

> 1. AI Spam. I categorized the inbound we got the other day from a job post. Out of 172 daily applicants, we got 22 that looked reasonably like a person, and 150 that were primarily AI generated messages. Which are pretty easy to spot because they're 500 words of tech jargon and rehashing the job description.

Is that not the ideal answer? Those that get to move forward are the one's that just happen to write a message that hits a non-descriptive sweet spot? A fucking Magic 8 Ball

Adding to this: we mostly focus on outbound, mostly using WorkAtAStartup or LinkedIn as a source.

The unspoken problem is that most people want way too much fucking money.

I don't think demand for engineers is necessarily that elastic - even if candidates were willing to accept less money, it would take a long time for companies to respond by hiring more and they might never match supply.

I've personally applied to several jobs offering less than $80k TC (government, nonprofits, startups) and heard nothing back. That's a small sample size because I'm currently employed and only looking at particularly interesting positions, but I'm also only applying to good fits, writing an individualized resume/cover letter, etc.

Anyways, I agree with GP that there seems to have been a breakdown of the old post job -> cold submit resume pipeline. You have to have some kind of an in these days.

My anecdotes last year and the year before was that when I was looking for my preferred jobs - full time positions at cloud consulting companies and a couple of product companies as an “architect”, I had a 100% response rate.

When I looked for my Plan B jobs - just regular old C# enterprise dev jobs looking for AWS experience paying less - crickets.

It’s seems harder to get interviews for jobs you are overqualified for than ones you are slightly under qualified for.

I have seen some YC startups offer absurdly low "fucking money" for the ask.

If you want a founding engineer for on-site only full-time work in the Bay Area at 50-80K and a fraction of a percent of equity, you deserve every volley of AI shitspam resume dumping that awaits your inbox.

And I think it was only two or three months ago that I saw a hiring post offering exactly that.

I have to double down on what I just said here especially these days because startups are dumping their early stage employees to keep them from actually keeping any of that equity more often than they should be.

You take all of the risk of early stage startup employment and as soon as the company gets long term traction you get shown the door.

I've seen this mostly in NY startups but have heard it's a wide trend.

So far we haven't had a candidate pass because of salary yet. Typically candidates looking for that stable $400k salary never even talk with startups, because they know that's only a FAANG position.

Out of curiosity, I looked at how much your company pays.

https://www.ycombinator.com/companies/omniai/jobs

This doesn’t come close to what a mid level enterprise CRUD developer could make in Atlanta.

I'm from Atlanta, and $175k is pretty square with ATL ranges[1]. It's not a Google salary, but startup comp is always more equity weighted.

[1] https://www.levels.fyi/t/software-engineer/locations/atlanta...

My bad, looking at the page from mobile it didn’t look like a salary range. I misinterpreted it. I just saw $125K.

That being said, I would hope that position is remote for $175K. I would have jumped on it when I was a pure software developer working in Atlanta. But working locally in SF for that amount?

If you can’t live comfortably on 175k in SF, you have a spending problem.

And according to the calculators I saw, it would be 46% more expensive than living in Atlanta. But we you can choose other metro areas that are not on the west coast for comparison where you could also make $175K as a senior

you guys like sunglasses

$145K is the median according to levels.fyi and 175K is around the 75th percentile.

According to salary.com, that’s the median for the highest level of IC in Atlanta.

https://www.salary.com/tools/salary-calculator/software-engi...

A person who takes a $175k/y but works 3.5h/d instead of 8h/d really costs the business $400k/y.

Your colleagues' network of high quality instant hires is only interested in switching to $400k/y opportunities.

This coin has many sides but it isn't complicated.

No people want the amount of money that the market will give them. If you can’t afford the inputs that are required for your business, that’s a you problem. Come back when you have enough funding to pique my interest with cash comp - not worthless “equity”

Interesting to me that 'working with AI' is using OpenAI's API.

Thats really working with someone elses AI, no?

FWIW I have used it quite a bit, but its not really the same thing as developing AI

(nevermind my usual rant that we shouldn't call any of this AI, but I digress)

I'd be fine calling myself a "AI Engineer" for having worked with TensorFlow, PyTorch and scikit-learn and having experience in running AI Models in production.

Nowadays, making requests to the API of a LLM provider is what makes an engineer a good fit for working at an AI company? Really?

> After three years of hearing about AI every day if you've never tested a single inference API then why are you applying to an AI startup.

Devil's advocate; why would I have made an API request if my employer has never used that service? Maybe that lack of interest on their part is why I'm trying to leave and get a job in a field that's of interest to me.

It begs the question: why not just post your job in your website and nowhere else (ie LinkedIn/Indeed)? That reduces the spam AND gives you a link to point people to.

I hear these complaints all the time from companies. To me if you’re leaving your jobs up at aggregators like LinkedIn, you have no right to complain (not singling you out btw, just a general thought)

Because maybe we'll get a good candidate. The cost is virtually free (30 minutes a day clearing an inbox). And once a week we get a decent application.

30 mins a day isn’t virtually free and last I checked it costs $$$ to post jobs on Linkedin

The irony of an AI company’s hiring woes due to too much AI.

Hey we process unstructured data so we're trying not to make things worse :')

I have a PhD from one of the best universities in the world in Machine Translation, I was training feed forward networks in 2014, built models in TensorFlow in 2016, founded a generative AI startup in 2019 and signed deals with huge consumer goods companies, debugged distributed training jobs on a cluster with thousands of A100s, and wrote PyTorch training pipelines for a financial ML trading signal that made money.

Yet, I am unqualified to join an AI startup because I’ve never made an API request to OpenAI

The attitude underlying this comment likely makes you a bad culture fit...

Make ur first API request to OpenAI today and who knows, rooting for ya!

This is a bad faith comment - if you had all that in your resume, they wouldnt disqualify you for not ‘sending an api request’.

Having a PhD vs. making that API call might ironically suggest you don’t emanate hands-on product thinking.

I know PhDs in data science and language technology who are not startup material because they’re too uncomfortable outside of MatLab/Jupyter/R.