Based on what I've seen/heard, they seem to follow a similar comp philosophy to Netflix -- pay top of market to get the best people.
But it's also important to note that only $300K of that is in cash. The other $600K is in profit participation, which could take years (maybe even a decade!) to be realized. It could also be worth $6M a year when it's realized.
But ultimately it's an investment of your time. Or to think of it another way, you're getting paid $900K a year but you're also investing $600K a year in OpenAI, which may end up being an amazing return or nothing at all, just like any startup investment.
Although with Sam at the helm, my guess is it will probably be worth more than $600K a year.
To be realistic ... 300K in the bay is barely livable if you're targeting a middle class life with kids and targeting retiring at 65.
First, ~46% of it is gone in taxes including federal tax (~25%), state tax (~8%), FICA (~4%), and sales tax on everything you eventually use the money for (~9%).
So that's 162K left. Not a lot to pay sky-high rents, car payments, insane medical bills despite insurance, lawyers to fight said bills, save up money for parental elderly care, save up money for yourself for retirement, etc.
And yeah, having kids on that money? Very difficult.
If you're not in the bay area, different story, it's a very nice income. But they probably won't give you that package if you're remote.
And if you're in the bay and not planning on having kids, it's an okay salary.
I guess the first part of that is like netflix, but as you know the second part is not since netflix is 100% cash unless you want to participate some portion of that in the employee stock plan.
Yes, the first part of the philosophy, pay top of market. Not the second part, where they employee gets to choose their level of exposure to investment risk. But Netflix is a public company, and OpenAI is not.
No, Netflix pays top of market like you said “in cash”. Paying “profit sharing” in a company that isn’t profitable isn’t nearly as valuable.
Edit:
I love this analogy
> Or to think of it another way, you're getting paid $900K a year but you're also investing $600K a year in OpenAI, which may end up being an amazing return or nothing at all, just like any startup investment.
Would someone invest 2/3rds of their compensation in one company? I know I diversify my RSUs within six months after getting them.
I feel like person you're replying to specifically implied that OpenAI does it differently in that aspect, and they were also high up at Netflix so is likely very familiar :)
> Would someone invest 2/3rds of their compensation in one company? I know I diversify my RSUs within six months after getting them.
I also sell my RSUs right away. But you're forgetting one thing -- if I make $600K somewhere else, I can't invest it in OpenAI. It's the opportunity to invest in a startup that you wouldn't get otherwise that is most valuable, if you think the startup will be valuable.
Beware of taxes though. If you leave the company before it's public, and the stock has increased in value, you have to pay capital gains taxes when exercising and there's the possibility that stock may end up worthless.
There have been many startup employees that have paid hundreds of thousands of dollars in taxes only to see their paper gains eventually disappear.
I recall a story someone told me a while ago. Software business that did local CoL/prevailing wages. Hired an intern one summer that was just running around in circles around the other, more senior devs. Useless to say they loved him and the next summer they tried to get him back, even offering a signing bonus for an internship (something they considered unheard of) but he was already at a large search engine company down in the Bay. You can guess the comp was probably already 3x what his previous job was offering. Of course, he wouldn't return.
There's a whole class of engineers were completely invisible to most companies, even if they are in the same "local market" [0][1] (Some use the term "dark matter devs" but I know it has another meaning [2]). These guys tend to fly under the radar quite a bit. If you are in a tier 2 market or company, your chances of attracting one are close to nil. Because they are extremely valuable, they don't interview a lot and tend to hop between companies where they know people (or get fast tracked internally).
FAANG companies have internship pipelines, with bonus for returning interns. These guys are off the market years before they even graduate.
I don't understand why people think sam has magic powers. He started one failed company and has no technical ability to make openai into anything more than it is.
yeah, Netflix is almost all cash in comparison. No clue what kind of liquidity OpenAI provides to cash out. Since it's "profit based" it actually seems even riskier, Microsoft could prevent OpenAI from showing any profit if they really want to
and if the info about the actual architecture of GPT-4 is legit, OpenAI might not have as much of an edge on the competition as they'd like people to believe. So that equity might not be worth quite as much as they think and explains why they so cagey about it in their paper on it. And why Sam Altman is calling for restrictions on research to slow down competitors
even if there was, while it might take a decade before IPO and market price discovery, it's almost expected that they would make a private tender offer sometime before that, letting employees realize some value of their shares long before 10 years is up.
It's always interesting to think about how someone somewhere is earning as much as you earn in a decade or lifetime in a single year. Also that you are that person for someone else.
A sole proprietor landscaper making $45-50K a year in California is paying $675 a year in annual registration fees just to keep his newish pickup truck on the road. Why newish? Because the people he's servicing trust a guy with a nicer work vehicle than a beaten down 30 year old Tacoma.
A $900k developer with the same pickup is also paying $675 a year.
Extrapolate this seemingly trivial example across literally EVERYTHING in life.
I do not find that interesting. If the goal is wealth redistribution, then either a marginal income/wealth (property)/sales tax accomplishes that.
We do not have marginal sales tax rates because it is not feasible.
Marginal income tax is feasible, and so it does exist in most places.
Marginal property/wealth tax is somewhere in the middle, given the difficulties in valuing thinly traded assets, and the tremendous effort required to appraise them all the time, over and over.
Don't forget, however, that the $45K landscaper probably pays zero in Federal taxes, while the 900K dev (assuming it was all cash, which it isn't) is losing a third of that to Fed taxes alone, and tons more to State and local.
I'm not saying the highly-paid developer is hurting, but the better question is why does it cost $675/year to register a vehicle in CA? Our income tax system is already highly progressive, so why do we have all these other stealth taxes that hurt low incomes the most.
I don't think that's a fair example. I'm not sure what it's like to run a small business in California. But the answer to that situation is progressive taxation, and California has one of the most progressive tax systems in the US. https://taxfoundation.org/which-states-have-most-progressive...
I am that person for myself, for my first year in the workforce.
I am also that person for myself when I was studying in college.
I think it is only interesting in the case when you are performing the exact same job, and yet the compensation varies by an order of magnitude or more.
I think the point is people earning $8/hr 40 hrs/wk 40 years = $665k total for a lifetime's work. Of course most people in the world earn much less than this.
One of the most compelling lyrics I've ever heard, that has stuck with me for about a decade now, is Chris Brown's, "I get what you get in 10 years in 2 days."
For people not living there but curious about trying it out: What's a reasonable range for "very well paid job in France"? Is it all cash or is there equity?
Other companies nickel and diming engineers think they are somehow getting something for free. All the highest performing companies in the world pay insane amounts to their programmers
Paying someone a lot doesn't magically make them a good engineer. You have to actually hire the best people, and there's only so many of them to go around.
I'm sympathetic to this argument, because I know both low-quality and high-quality engineers at most FAANGs, but in principle isn't offering a lot of money a good way to attract talent?
In the case of OpenAI you also have interesting tech and a brand that will massively accelerate your career if you want to stay in that field. So while yeah, you have to hire the best people; and OpenAI like everyone else will be paying $LOTS to a few useless engineers in the mix, I think "$900K and everyone knows it" is a pretty good substitute for talent-spotting, which anyway can't be bought.
Paying someone more also doesn’t mean you can charge your customers more. A budget is a budget. There’s a reason McDonald’s pays employees differently in India vs NYC, the work maybe the same but the market is different
Meta cut back on everything: microkitchens, wellness cash, assigned->shared desks, WfH->RTO, and looking for excuses to lay people off. I wouldn't take a job at Google, Amazon, Apple, or Meta. Netflix and Microsoft seem doable. Target and Costco are also decent non-tech companies.
I think this is often, but not always true. I think it's not true (and rational for employers) if they:
* have a set of interview questions that help clarify they have the "right" kind of person
* have a performance review process that ensures that hired individuals continue to engage in ways that are understood to be most useful to the company; i.e. that validate (or invalidate, as the case may be) hiring practices
The unfortunate reality is that those elite circles usually do have the best among them. And having worked in FAANGs, there is some aspect of meritocracy, the code doesnt lie
Hmm I’ve worked at “elite” startups and “elite” big cos, all jobs of which I’ve gotten by directly applying. At several points have exceeded 7 figures too.
Talent is not universal though. A great product/team/workplace can turn unmotivated middling knowledge people into heroes. Not only is massively more worth to raise talent in-house, heterogenous teams are way more healthy. Early Googlers wrote litanies on this.
They can pay it, it sounds good for investors/general public, but if you take away the brand/hype fapping on "we only have the best" would sounds like a red flag to an experienced engineer.
Can we define talent. Also context matters. Some companies have lower "talent density" because their size requires some level of operational headcount.
I know more than half a dozen extremely top of their field for sure engineers who work there. Even having them would be an insane concentration of talent imo.
> The tax filings also reveal previously unreported details about one of the most valuable and well-known technology ventures operating today, including the level of investment by Reid Hoffman, free Teslas for early OpenAI engineers and the skyrocketing computing bill that may have prompted it to take a $1 billion investment from Microsoft.
New hires' comp is much higher than existing employees', especially if you've hit your cliff. 7 figures for E6 can happen if you joined recently, have good counter-offers, and negotiate. It's not super uncommon but it's also not the median E6 comp.
Some made up new equity class that isn't actually equity, and caps your upside? Are we seeing a new level of startup management greed? That smells pretty scammy.
Common advice is to value earlier stage equity at zero anyway, so how the heck are people credibly calling that worth 600k...
There are a handful of unicorns / decacorns that went poof and the BS equity turned out to indeed be BS. That’s why they have to do $300k base salary, because that’s low end of total comp at competing mega caps, without it they couldn’t sign anybody from there.
OpenAI will probably end up like UberATG where Microsoft bails out a handful of $10m-plus equity packages, some engineers transfer into MS and get partial recognition, and a bunch of engineers get laid off or sent to a competitor for basically zero. Good chance of this happening when sama has to scale up aggressively and flops.
Unpopular opinion: Paying more does not get you better code, or faster code, or cleaner code. It does get you valuable things, but they're unrelated to code. I'll get to that.
Some of the best all-around programmers I've met, and also the best programmers at some specific metric (rapid development, good at unraveling complex problems, good at big-picture stuff, good at architecture, good at maths/algorithms, etc.) make market rate or below market rate. Probably because they either (A) don't know their value (B) didn't go to a top CS school (C) value things other than raw income (D) aren't good at playing the political game at work (E) aren't good at self-promotion.
What paying high salaries does get you is... the widely-publicized knowledge that you pay high salaries. Sites like levels.fyi and HN will viralize this knowledge. Like your neighbor who has a shiny new BMW M3 in their driveway, you assume that they must be doing well, and must be a good investment.
OpenAI PPUs are no different than stock in a growing-but-unprofitable company like old Amazon. You don't have to wait for dividends; you can sell your PPUs to speculators who expect returns in the future.
> It’s important to reiterate that the PPUs inherently are not redeemable for value if OpenAI does not turn a profit. That said, there are investors that have been willing to pay for these PPUs and that’s where the value being told to candidates is derived from.
> In conversations with recruiters we’ve heard from some candidates that OpenAI is communicating that they don’t expect to turn a profit until they reach their mission of Artificial General Intelligence (AGI). As with any startup, there are risks associated with equity and it’s important to consider the possibility that equity can be worth nothing.
This is weird. They don’t expect to be profitable until AGI exists? Does the monetary system we have today even matter at that point?
If OpenAI has a monopoly on AGI within the current neoliberal capitalist framework they'll make trillions before the system collapses. It's like Aramco but bigger.
Do the PPU's create an alignment issue? I mean, theoretically these folks are supposed to be working towards AGI, but doing it safely... but I can imagine that an extra 500K+ is a pretty big incentive to hurry up and create something that can help you make a profit so that they can get their payout. Of course, with the outside investment from Microsoft, maybe this is somewhat moot anyways as MSFT definitely has a very strong incentive to get OpenAI to build fast.
Engineers don't control the strategy. If Altman decides to reinvest all free cash flow into growth and safety instead of profits (which is not unusual in tech) the employees can't/won't stop him.
Maybe not, but if as others have mentioned in the thread that there is very high talent density and these are top folks, then surely engineers have some say in the strategy -- if only through mass defection and quitting, right?
Their job postings are (were?) certainly...a thing. I can't find the page (maybe I hallucinated it), but I remember reading a line that was essentially "we only want famous ML/AI researchers and engineers, others need not apply". When I read it a few months ago I said "yeah, that makes sense, everyone wants to work for you right now" and closed the page. It looks like they are casting a wider net now, though.
$600K of that is “equity” in a private company. You never know when the company will go public or with what valuation.
I can’t sell equity in a private company to exchange for goods and services like I can stock that gets deposited in my brokerage account every six months.
There are exceptions right? Stripe isn't public yet I'm pretty sure there are ways to sell the stock that's accrued. I remember something about them opening up a way to sell your shares and since I get pinged on LinkedIn every so often about it I'm assuming it's a thing.
Only at a deep “we buy ugly houses” type of discount and only if the company allows it and if there is a market for it. Like I said earlier, I know the value of my “equity” in a publicly held company:
=GOOGLEFINANCE("AMZN", "price")* (number of shares).
I can log into Fidelity and sell my shares during hours when the market is open and not at a discount.
If it is “equity” in a private company, any employees can certainly sell that stock to a venture buyer on the secondary market. However the share price will likely be discounted.
What type of market do you think there was to buy my options in the 60 person startup that no one ever heard of at my previous company? I was paid in cash at market value so it wasn’t a big deal - and “market” for your standard enterprise CRUD developer architect was nothing to write home about compared to tech company salaries.
They do hire remote in some cases, but normally only for up to a year and only if it's someone they really want. They hope that the person will be willing to move after working there for a year, I guess.
Do you guys know some good companies that offer interesting remote ML jobs? The package is competitive for sure. I don't care that much about the pay, more about interesting stuff.
Same, and the base salaries from their actual job listings were mostly 200-270K per year. Definitely not a bad salary, but not worth it given the cost of SF living.
I mean, Netflix-style hunger-games management might be their thing, and if I got in there at $900K (I did apply!) -- then sure, I'd be working nights and weekends just to shine.
But even at infinite comp, above a certain size there will be dead weight.
Is Netflix really that bad? They aggressively state they quickly fire people. But I've also seen reports that their turnover rate is low, and they don't have a PIP culture. It looks to me like people are reasonably happy there for the most part.
There’s been a couple times in my career where my income needs were low enough that for one more raise I could afford to hire an assistant. For that kind of money I could hire several. And an apprentice.
I've never looked into their interview process, but considering what they have created I would guess that their coding problems go way beyond leetcode.
Hard part is getting an interview. It helps a lot to know someone already inside, and this is pretty much standard for all companies regardless of prestige.
If you're really well-known, they'll come to you. Otherwise, get an insider referral.
MTS is usually what a recent college graduate gets promoted to, once they've passed their sort of 1 year trial period. You can +/- that timeline about a year, but more or less that's what it is, at least from what I've seen in SF.
The [dead] comment replying to you doesn’t seem to understand that “Staff Engineer” is not typically the same as “Member of the Technical Staff.” Not sure why they had to express that misunderstanding in such a tone. Usually it’s better to approach a situation with curiosity when you don’t know what you’re talking about, rather than anger.
"yes, i produce automated plagiarism which can only function on the backs of exploited Kenyan workers, encodes and furthers every systemic bias imaginable and endangers society -- but at least I am paid well"
That's getting close to some Nuremberg level shit. Not yet there of course but consider when the next pandemic rolls around and the anti science propaganda will be produced by these how many will die?
Unpopular opinion: OpenAI is severely underpaying these folks at 900k considering so many asset managers / PMs in finance basically hug S&P 500 and get 1.5m+
Another unpopular opinion: If I think about the reasons why asset managers PMs in finance are overpaid, it's because they created elaborate rent seeking structure in society and industry, with disequilibrium in penalty vs reward. Same thing for lawyers in big firms.
I would argue the engineers in big techs also have benefited from the rent seeking structure. The only reason why they're not enjoying bigger payout is because the actual structure was created by the big tech companies, and not through coordinated effort by engineers. As a result, they reap less reward as they're just highly paid henchmen, and not the bosses.
Yes, but this is just because tech industry people are far more susceptible to fantasies and bullshit than those in finance. In fact, they seem to largely enjoy the fantasies and bullshit, as this whole AGI rigamarole has shown. They’ll take a pay cut because the enjoy the make-believe over the cold, naked logic of market movements.
It's too bad about their outdated stance on remote work. They could pay $2m and many of the best engineers still wouldn't consider working there on that point alone.
There is little maths to show off "AI skills", then there are many scammers.
If there is so much money, which is where the "absurd" is, better target people with validated maths PHDs. Moreover, those researchers will need access to super computers in order to test at scale their work (because it all about stability at scale), and that's a limiting parameter.
Additionally, "AI" is just a set of empirical algorithms attempting to mimic human cognition. The trick is to find mathematical stability (production, training, etc) for large scale simulated connectomes: I am still surprised there is still not a "sleep" regime for those articial connectomes, which we know is critical for brain stability, maybe it is a part of the new "training" regimes.
I am still waiting for the educational material which will let people picture a connectome based on the latest "transformers". Because, in the end, nobody understands anything about this, it is beyond us. It will be hard to tell apart the scammers looking to steal fundings from the others.
Based on what I've seen/heard, they seem to follow a similar comp philosophy to Netflix -- pay top of market to get the best people.
But it's also important to note that only $300K of that is in cash. The other $600K is in profit participation, which could take years (maybe even a decade!) to be realized. It could also be worth $6M a year when it's realized.
But ultimately it's an investment of your time. Or to think of it another way, you're getting paid $900K a year but you're also investing $600K a year in OpenAI, which may end up being an amazing return or nothing at all, just like any startup investment.
Although with Sam at the helm, my guess is it will probably be worth more than $600K a year.
Honestly, if you take 300k home, the risk isn't that high. But the reward is insane.
That person could probably be making 500k+ at a FANG with less variance in the stock component.
44 replies →
To be realistic ... 300K in the bay is barely livable if you're targeting a middle class life with kids and targeting retiring at 65.
First, ~46% of it is gone in taxes including federal tax (~25%), state tax (~8%), FICA (~4%), and sales tax on everything you eventually use the money for (~9%).
So that's 162K left. Not a lot to pay sky-high rents, car payments, insane medical bills despite insurance, lawyers to fight said bills, save up money for parental elderly care, save up money for yourself for retirement, etc.
And yeah, having kids on that money? Very difficult.
If you're not in the bay area, different story, it's a very nice income. But they probably won't give you that package if you're remote.
And if you're in the bay and not planning on having kids, it's an okay salary.
62 replies →
I guess the first part of that is like netflix, but as you know the second part is not since netflix is 100% cash unless you want to participate some portion of that in the employee stock plan.
Yes, the first part of the philosophy, pay top of market. Not the second part, where they employee gets to choose their level of exposure to investment risk. But Netflix is a public company, and OpenAI is not.
4 replies →
The part about paying top of market to get the best talent is the same. The breakdown of comp is not.
1 reply →
No, Netflix pays top of market like you said “in cash”. Paying “profit sharing” in a company that isn’t profitable isn’t nearly as valuable.
Edit:
I love this analogy
> Or to think of it another way, you're getting paid $900K a year but you're also investing $600K a year in OpenAI, which may end up being an amazing return or nothing at all, just like any startup investment.
Would someone invest 2/3rds of their compensation in one company? I know I diversify my RSUs within six months after getting them.
I feel like person you're replying to specifically implied that OpenAI does it differently in that aspect, and they were also high up at Netflix so is likely very familiar :)
> Would someone invest 2/3rds of their compensation in one company? I know I diversify my RSUs within six months after getting them.
I also sell my RSUs right away. But you're forgetting one thing -- if I make $600K somewhere else, I can't invest it in OpenAI. It's the opportunity to invest in a startup that you wouldn't get otherwise that is most valuable, if you think the startup will be valuable.
1 reply →
Beware of taxes though. If you leave the company before it's public, and the stock has increased in value, you have to pay capital gains taxes when exercising and there's the possibility that stock may end up worthless.
There have been many startup employees that have paid hundreds of thousands of dollars in taxes only to see their paper gains eventually disappear.
I recall a story someone told me a while ago. Software business that did local CoL/prevailing wages. Hired an intern one summer that was just running around in circles around the other, more senior devs. Useless to say they loved him and the next summer they tried to get him back, even offering a signing bonus for an internship (something they considered unheard of) but he was already at a large search engine company down in the Bay. You can guess the comp was probably already 3x what his previous job was offering. Of course, he wouldn't return.
There's a whole class of engineers were completely invisible to most companies, even if they are in the same "local market" [0][1] (Some use the term "dark matter devs" but I know it has another meaning [2]). These guys tend to fly under the radar quite a bit. If you are in a tier 2 market or company, your chances of attracting one are close to nil. Because they are extremely valuable, they don't interview a lot and tend to hop between companies where they know people (or get fast tracked internally).
FAANG companies have internship pipelines, with bonus for returning interns. These guys are off the market years before they even graduate.
[0] https://blog.pragmaticengineer.com/software-engineering-sala...
[1] http://danluu.com/bimodal-compensation/
[2] https://www.hanselman.com/blog/dark-matter-developers-the-un...
This is why Sam wants regulation.
All this money spent, yet if the open source world can catch up, it'll all be for naught.
They need a world where everyone sits atop OpenAI, not a thousand different startups with their own models.
OpenAI might pay its engineers $900k, but an AI startup founder can easily get to a $5M, $10M valuation in under a year.
> you're getting paid $900K a year but you're also investing $600K a year in OpenAI
It might sound simple, but I've never viewed startup equity like that; thanks for the slightly different perspective.
I don't understand why people think sam has magic powers. He started one failed company and has no technical ability to make openai into anything more than it is.
yeah, Netflix is almost all cash in comparison. No clue what kind of liquidity OpenAI provides to cash out. Since it's "profit based" it actually seems even riskier, Microsoft could prevent OpenAI from showing any profit if they really want to
and if the info about the actual architecture of GPT-4 is legit, OpenAI might not have as much of an edge on the competition as they'd like people to believe. So that equity might not be worth quite as much as they think and explains why they so cagey about it in their paper on it. And why Sam Altman is calling for restrictions on research to slow down competitors
https://twitter.com/soumithchintala/status/16712671501017210...
Open AI does not make profits though
Although with Sam at the helm, my guess is it will probably be worth more than $600K a year.
or not . I would rather work for facebook with profit participation. zuck seems to know how to make a profit.
I would take it. There are some interesting things to work on.
sure but it looks like you can sell PPU's to investors?
no "right to first refusal" or some transfer restriction?
even if there was, while it might take a decade before IPO and market price discovery, it's almost expected that they would make a private tender offer sometime before that, letting employees realize some value of their shares long before 10 years is up.
1 reply →
It's always interesting to think about how someone somewhere is earning as much as you earn in a decade or lifetime in a single year. Also that you are that person for someone else.
What I find more interesting is this:
A sole proprietor landscaper making $45-50K a year in California is paying $675 a year in annual registration fees just to keep his newish pickup truck on the road. Why newish? Because the people he's servicing trust a guy with a nicer work vehicle than a beaten down 30 year old Tacoma.
A $900k developer with the same pickup is also paying $675 a year.
Extrapolate this seemingly trivial example across literally EVERYTHING in life.
I do not find that interesting. If the goal is wealth redistribution, then either a marginal income/wealth (property)/sales tax accomplishes that.
We do not have marginal sales tax rates because it is not feasible.
Marginal income tax is feasible, and so it does exist in most places.
Marginal property/wealth tax is somewhere in the middle, given the difficulties in valuing thinly traded assets, and the tremendous effort required to appraise them all the time, over and over.
7 replies →
Yeah crazy. A gallon of milk costs the same if you’re a landscaper or a tech bro. Really makes ya think.
2 replies →
Don't forget, however, that the $45K landscaper probably pays zero in Federal taxes, while the 900K dev (assuming it was all cash, which it isn't) is losing a third of that to Fed taxes alone, and tons more to State and local.
I'm not saying the highly-paid developer is hurting, but the better question is why does it cost $675/year to register a vehicle in CA? Our income tax system is already highly progressive, so why do we have all these other stealth taxes that hurt low incomes the most.
I don't understand your point - is it that people should pay different amount for identical amount of work based on how much they themselves are paid?
8 replies →
I don't think that's a fair example. I'm not sure what it's like to run a small business in California. But the answer to that situation is progressive taxation, and California has one of the most progressive tax systems in the US. https://taxfoundation.org/which-states-have-most-progressive...
1 reply →
I find it more interesting that the sole proprietor landscaper doesn’t just become a developer and make $900k a year.
Extrapolate this to literally anyone making less than a developer.
5 replies →
The sole proprietor is also able to get a tax deduction on his truck that the developer can't.
2 replies →
which is why billionaires are an indictment to our system.
3 replies →
Is it interesting?
I am that person for myself, for my first year in the workforce.
I am also that person for myself when I was studying in college.
I think it is only interesting in the case when you are performing the exact same job, and yet the compensation varies by an order of magnitude or more.
I think the point is people earning $8/hr 40 hrs/wk 40 years = $665k total for a lifetime's work. Of course most people in the world earn much less than this.
1 reply →
One of the most compelling lyrics I've ever heard, that has stuck with me for about a decade now, is Chris Brown's, "I get what you get in 10 years in 2 days."
For me, on the same track, I'd have to go with Busta's verse when he's all "cuz it doesn't matter cuz I'm gonna da da da da"
(I jest. I just really like that song's beat & Busta's verse in terms of sheer speed & enunciation & crispness in staying on beat.)
1 reply →
> you are that person for someone else
Especially if you're American. Gallup [1] reports:
> The median annual household income worldwide is $9,733, and the median per-capita household income is $2,920
Even among developed nations, Americans get paid more.
[1] https://news.gallup.com/poll/166211/worldwide-median-househo...
This is a good point.
I look at 900k$ and think wow from the perspective of my very well paid job in France.
Getting 900k$ a year would bring great things to me and probably make me retire earlier. This will not be life-changing though.
And then I think about people who earn 5% of what I do and for them multiplying their income by two would probably be truly life changing for them.
And then these 900k$ do not look that wow anymore.
I'd call "may retire, or work on whatever you want for the rest of your life, after a few years of saving" life-changing.
2 replies →
For people not living there but curious about trying it out: What's a reasonable range for "very well paid job in France"? Is it all cash or is there equity?
4 replies →
Are you here to just brag? Also the dollar sign goes on the other side.
4 replies →
I find it interesting too. Also interesting to see how I measure with regard to my parents.
Other companies nickel and diming engineers think they are somehow getting something for free. All the highest performing companies in the world pay insane amounts to their programmers
Paying someone a lot doesn't magically make them a good engineer. You have to actually hire the best people, and there's only so many of them to go around.
I'm sympathetic to this argument, because I know both low-quality and high-quality engineers at most FAANGs, but in principle isn't offering a lot of money a good way to attract talent?
In the case of OpenAI you also have interesting tech and a brand that will massively accelerate your career if you want to stay in that field. So while yeah, you have to hire the best people; and OpenAI like everyone else will be paying $LOTS to a few useless engineers in the mix, I think "$900K and everyone knows it" is a pretty good substitute for talent-spotting, which anyway can't be bought.
11 replies →
Right you are correct. But true talent on any random team is easily worth 3x the average salary on that team. Im speaking about programmers here
9 replies →
Paying someone more also doesn’t mean you can charge your customers more. A budget is a budget. There’s a reason McDonald’s pays employees differently in India vs NYC, the work maybe the same but the market is different
Meta cut back on everything: microkitchens, wellness cash, assigned->shared desks, WfH->RTO, and looking for excuses to lay people off. I wouldn't take a job at Google, Amazon, Apple, or Meta. Netflix and Microsoft seem doable. Target and Costco are also decent non-tech companies.
overpaying is going to attract the wrong people
I think this is often, but not always true. I think it's not true (and rational for employers) if they:
* have a set of interview questions that help clarify they have the "right" kind of person
* have a performance review process that ensures that hired individuals continue to engage in ways that are understood to be most useful to the company; i.e. that validate (or invalidate, as the case may be) hiring practices
Who are these “wrong people”? The ones who “don’t believe in the mission and are just working to get paid”?
Only when overfunded with VC money.
At that point, they are not hiring you, me or anyone who directly applies themselves via the job form.
They are hiring their friends from elite circles.
The unfortunate reality is that those elite circles usually do have the best among them. And having worked in FAANGs, there is some aspect of meritocracy, the code doesnt lie
8 replies →
Hmm I’ve worked at “elite” startups and “elite” big cos, all jobs of which I’ve gotten by directly applying. At several points have exceeded 7 figures too.
Sam Altman said something to the effect of 'OpenAI has the highest talent density of any company'.
OpenAI is modest in employee count and extreme in employee talent, with its employees rewarded commensurately.
Talent is not universal though. A great product/team/workplace can turn unmotivated middling knowledge people into heroes. Not only is massively more worth to raise talent in-house, heterogenous teams are way more healthy. Early Googlers wrote litanies on this.
They can pay it, it sounds good for investors/general public, but if you take away the brand/hype fapping on "we only have the best" would sounds like a red flag to an experienced engineer.
Can we define talent. Also context matters. Some companies have lower "talent density" because their size requires some level of operational headcount.
I know more than half a dozen extremely top of their field for sure engineers who work there. Even having them would be an insane concentration of talent imo.
Maybe for its size but I bet there are lots of smaller companies with higher talent density, e.g. Oxide Computer.
> The tax filings also reveal previously unreported details about one of the most valuable and well-known technology ventures operating today, including the level of investment by Reid Hoffman, free Teslas for early OpenAI engineers and the skyrocketing computing bill that may have prompted it to take a $1 billion investment from Microsoft.
https://techcrunch.com/2023/05/17/elon-musk-used-to-say-he-p...
A ~$100K signing bonus sounds kind of small for someone making $900K per year.
A 100k _signing_ bonus isn't small by any measure whatsoever.
1 reply →
How? It's over 10% of the yearly salary. That seems in-line if not better than most signing bonuses I've seen.
A signing bonus the size of an average software engineer salary here...and I'm in a very wealthy country
Because the PPU equity isn’t worth anything until they’re profitable.
7-figure total comp is common for E6 and very common for E7 engineers at Meta. A stock being a significant part of course.
The median OAI employee with a 900k comp is probably L5, not L6 or L7
Why does levels.fyi say US$542,547 for E6? I count only 6 figures.
New hires' comp is much higher than existing employees', especially if you've hit your cliff. 7 figures for E6 can happen if you joined recently, have good counter-offers, and negotiate. It's not super uncommon but it's also not the median E6 comp.
1 reply →
Some made up new equity class that isn't actually equity, and caps your upside? Are we seeing a new level of startup management greed? That smells pretty scammy.
Common advice is to value earlier stage equity at zero anyway, so how the heck are people credibly calling that worth 600k...
There are a handful of unicorns / decacorns that went poof and the BS equity turned out to indeed be BS. That’s why they have to do $300k base salary, because that’s low end of total comp at competing mega caps, without it they couldn’t sign anybody from there.
OpenAI will probably end up like UberATG where Microsoft bails out a handful of $10m-plus equity packages, some engineers transfer into MS and get partial recognition, and a bunch of engineers get laid off or sent to a competitor for basically zero. Good chance of this happening when sama has to scale up aggressively and flops.
Unpopular opinion: Paying more does not get you better code, or faster code, or cleaner code. It does get you valuable things, but they're unrelated to code. I'll get to that.
Some of the best all-around programmers I've met, and also the best programmers at some specific metric (rapid development, good at unraveling complex problems, good at big-picture stuff, good at architecture, good at maths/algorithms, etc.) make market rate or below market rate. Probably because they either (A) don't know their value (B) didn't go to a top CS school (C) value things other than raw income (D) aren't good at playing the political game at work (E) aren't good at self-promotion.
What paying high salaries does get you is... the widely-publicized knowledge that you pay high salaries. Sites like levels.fyi and HN will viralize this knowledge. Like your neighbor who has a shiny new BMW M3 in their driveway, you assume that they must be doing well, and must be a good investment.
300k salary + 600k stock
Not stock PPUs which are only worth anything if they turn a profit and according to OpenAI that isn’t happening until AGI.
https://www.levels.fyi/blog/openai-compensation.html
OpenAI PPUs are no different than stock in a growing-but-unprofitable company like old Amazon. You don't have to wait for dividends; you can sell your PPUs to speculators who expect returns in the future.
5 replies →
Wonder if we'll see many IPOs from any companies not turning a profit anytime soon
UBI will just be everyone getting paid dividends from OpenAI stock
Instead of traditional equity, OpenAI issues PPUs which represent a share of the profits. More details on how it works here: https://www.levels.fyi/blog/openai-compensation.html
> It’s important to reiterate that the PPUs inherently are not redeemable for value if OpenAI does not turn a profit. That said, there are investors that have been willing to pay for these PPUs and that’s where the value being told to candidates is derived from.
> In conversations with recruiters we’ve heard from some candidates that OpenAI is communicating that they don’t expect to turn a profit until they reach their mission of Artificial General Intelligence (AGI). As with any startup, there are risks associated with equity and it’s important to consider the possibility that equity can be worth nothing.
This is weird. They don’t expect to be profitable until AGI exists? Does the monetary system we have today even matter at that point?
If OpenAI has a monopoly on AGI within the current neoliberal capitalist framework they'll make trillions before the system collapses. It's like Aramco but bigger.
2 replies →
Every job I checked has a cap of $370k TC https://openai.com/careers/search
That's not TC, it's just the salary. From the job listing: "Total compensation also includes generous equity and benefits"
Do the PPU's create an alignment issue? I mean, theoretically these folks are supposed to be working towards AGI, but doing it safely... but I can imagine that an extra 500K+ is a pretty big incentive to hurry up and create something that can help you make a profit so that they can get their payout. Of course, with the outside investment from Microsoft, maybe this is somewhat moot anyways as MSFT definitely has a very strong incentive to get OpenAI to build fast.
Engineers don't control the strategy. If Altman decides to reinvest all free cash flow into growth and safety instead of profits (which is not unusual in tech) the employees can't/won't stop him.
Maybe not, but if as others have mentioned in the thread that there is very high talent density and these are top folks, then surely engineers have some say in the strategy -- if only through mass defection and quitting, right?
1 reply →
Their job postings are (were?) certainly...a thing. I can't find the page (maybe I hallucinated it), but I remember reading a line that was essentially "we only want famous ML/AI researchers and engineers, others need not apply". When I read it a few months ago I said "yeah, that makes sense, everyone wants to work for you right now" and closed the page. It looks like they are casting a wider net now, though.
$600K of that is “equity” in a private company. You never know when the company will go public or with what valuation.
I can’t sell equity in a private company to exchange for goods and services like I can stock that gets deposited in my brokerage account every six months.
There are exceptions right? Stripe isn't public yet I'm pretty sure there are ways to sell the stock that's accrued. I remember something about them opening up a way to sell your shares and since I get pinged on LinkedIn every so often about it I'm assuming it's a thing.
Only at a deep “we buy ugly houses” type of discount and only if the company allows it and if there is a market for it. Like I said earlier, I know the value of my “equity” in a publicly held company:
=GOOGLEFINANCE("AMZN", "price")* (number of shares).
I can log into Fidelity and sell my shares during hours when the market is open and not at a discount.
3 replies →
If it is “equity” in a private company, any employees can certainly sell that stock to a venture buyer on the secondary market. However the share price will likely be discounted.
Not “any employee” - the company has to allow it.
What type of market do you think there was to buy my options in the 60 person startup that no one ever heard of at my previous company? I was paid in cash at market value so it wasn’t a big deal - and “market” for your standard enterprise CRUD developer architect was nothing to write home about compared to tech company salaries.
You can absolutely sell the PPUs in OpenAI, you will not struggle to find a buyer.
At the full value of $600K?
Looks like they don't hire remote, otherwise I would absolutely be applying right now...no way am I moving to SF.
(Not that I think I would have a realistic chance, since I'm sure they're getting the best of the best applicants right now. But still.)
They do hire remote in some cases, but normally only for up to a year and only if it's someone they really want. They hope that the person will be willing to move after working there for a year, I guess.
Do you guys know some good companies that offer interesting remote ML jobs? The package is competitive for sure. I don't care that much about the pay, more about interesting stuff.
Same, and the base salaries from their actual job listings were mostly 200-270K per year. Definitely not a bad salary, but not worth it given the cost of SF living.
Is that how they’ll build a moat? Good luck I guess.
So do you gotta grind leetcode for this or what
you grinding leetcode is how they built this in the first place
If by Leetcode, you mean publish highly-cited ML papers, then yes.
Nah, you have to write cute little LLMs that grind leetcode for you...
Guessing they're looking at degrees and/or experience.
According to another poster, be a famous ML engineer…
What should I know to get a job there?
I rather suspect, they'll know you.
Now if only they would hire remote...
But it's all in paper clips...
You better perform with that pay
Or...?
I mean, Netflix-style hunger-games management might be their thing, and if I got in there at $900K (I did apply!) -- then sure, I'd be working nights and weekends just to shine.
But even at infinite comp, above a certain size there will be dead weight.
Is Netflix really that bad? They aggressively state they quickly fire people. But I've also seen reports that their turnover rate is low, and they don't have a PIP culture. It looks to me like people are reasonably happy there for the most part.
1 reply →
There’s been a couple times in my career where my income needs were low enough that for one more raise I could afford to hire an assistant. For that kind of money I could hire several. And an apprentice.
How do you objectively measure the performance of an engineer?
Genuinely curious.
Now the problem is how to get into this kind of companies. Is LC hard mandatory for junior-senior (below lead) IC job?
I've never looked into their interview process, but considering what they have created I would guess that their coding problems go way beyond leetcode.
Definitely. But do they hire people who can immediately start groking those problems or grow them? I guess they are also picky about colleges too?
Probably not true for their front-end positions.
Hard part is getting an interview. It helps a lot to know someone already inside, and this is pretty much standard for all companies regardless of prestige.
If you're really well-known, they'll come to you. Otherwise, get an insider referral.
True. I need to work on my connections.
LC hard shouldn't be needed for almost any companies except ones which are delusional. LC hard are there as practice for programming contests.
Thanks. I did hear that contests can be way more difficult than LC hard. But I'm struggling with medium anyway.
Isn't it for member of technical staff? It's a high level position.
MTS is usually what a recent college graduate gets promoted to, once they've passed their sort of 1 year trial period. You can +/- that timeline about a year, but more or less that's what it is, at least from what I've seen in SF.
The [dead] comment replying to you doesn’t seem to understand that “Staff Engineer” is not typically the same as “Member of the Technical Staff.” Not sure why they had to express that misunderstanding in such a tone. Usually it’s better to approach a situation with curiosity when you don’t know what you’re talking about, rather than anger.
[dead]
All engineers and researchers, even junior, are "Member of Technical Staff" at OAI
Everyone is a member of technical staff.
Is it? At some companies that’s the baseline title for engineers, data scientists etc.
"yes, i produce automated plagiarism which can only function on the backs of exploited Kenyan workers, encodes and furthers every systemic bias imaginable and endangers society -- but at least I am paid well"
That's getting close to some Nuremberg level shit. Not yet there of course but consider when the next pandemic rolls around and the anti science propaganda will be produced by these how many will die?
We know manually produced anti vaxx propaganda killed about 300 000 people in the United States (source: https://www.npr.org/sections/health-shots/2022/05/13/1098071... ) , what it'll be next time?
[flagged]
Unpopular opinion: OpenAI is severely underpaying these folks at 900k considering so many asset managers / PMs in finance basically hug S&P 500 and get 1.5m+
Another unpopular opinion: If I think about the reasons why asset managers PMs in finance are overpaid, it's because they created elaborate rent seeking structure in society and industry, with disequilibrium in penalty vs reward. Same thing for lawyers in big firms.
I would argue the engineers in big techs also have benefited from the rent seeking structure. The only reason why they're not enjoying bigger payout is because the actual structure was created by the big tech companies, and not through coordinated effort by engineers. As a result, they reap less reward as they're just highly paid henchmen, and not the bosses.
Yes, but this is just because tech industry people are far more susceptible to fantasies and bullshit than those in finance. In fact, they seem to largely enjoy the fantasies and bullshit, as this whole AGI rigamarole has shown. They’ll take a pay cut because the enjoy the make-believe over the cold, naked logic of market movements.
Another way to think about is that finance attracts people who care about money, tech does not.
2 replies →
Any engineer holding equity before November of 2022 is surely going to be worth > 10M
It's too bad about their outdated stance on remote work. They could pay $2m and many of the best engineers still wouldn't consider working there on that point alone.
Oh no, I will let Sam know right away!
[dead]
It’s incredible that in some part of the world the programmers are called “nerds” but in San Francisco are millionaires.
They're not mutually exclusive, I embrace the term, albeit I'm not a millionaire!
more like tens or hundreds of millions-aires
This is excessive. It is high-school level (in the worst case scenario 1 or 2 college/uni years).
Yeah, a bit higher than the minimum wage.
It seems to be inflated wages not based on skills but on technical fashion trends.
Mafia?
I am very confused by your comment. Are you claiming that $900k median compensation is only a bit higher than minimum wage?
Or are you claiming that some of the smartest AI researchers in the world are only slightly more qualified than the average high-schooler?
Either claim is roughly equally absurd.
There is little maths to show off "AI skills", then there are many scammers.
If there is so much money, which is where the "absurd" is, better target people with validated maths PHDs. Moreover, those researchers will need access to super computers in order to test at scale their work (because it all about stability at scale), and that's a limiting parameter.
Additionally, "AI" is just a set of empirical algorithms attempting to mimic human cognition. The trick is to find mathematical stability (production, training, etc) for large scale simulated connectomes: I am still surprised there is still not a "sleep" regime for those articial connectomes, which we know is critical for brain stability, maybe it is a part of the new "training" regimes.
I am still waiting for the educational material which will let people picture a connectome based on the latest "transformers". Because, in the end, nobody understands anything about this, it is beyond us. It will be hard to tell apart the scammers looking to steal fundings from the others.