I always love these posts. Great job to the interns. Building a tokenizer with a handcrafted automaton instead of a regex sounds like a cool project. And I found it interesting that Jane Street has a frontend for interacting with different AI services and LLMs. I'm curious how that's used.
But to be honest, what surprised me most about that was that Jane Street cares about counting the token usage in its API calls. I can see why they would care if they were interacting with AI models at very large scales, but are they? They have billions of dollars, so it struck me as odd to see such concern for efficiency in what must be a very small cost center.
I wonder if the motivation is more about providing feedback to the users of the frontend tool, as they write their prompts, so they can gain an intuition for what makes an efficiently tokenized prompt? (So that when the day comes that they are running these prompts at scale, the token sizes really matter). Or maybe I'm imagining "prompts" wrong, and these are not just a few paragraphs of text, but actually 20,000+ tokens at a time, including data like e.g. pasted CSVs of trading history for the LLM to analyze?
Token counting is importing when you are injecting fetched data into the prompt to make sure you don't overflow the prompt size (e.g. in retrieval augmented generation). You want to give the LLM as many facts as will fit in the prompt to improve the quality of its response. So even with billions of dollars... token counting is a thing.
>They have billions of dollars, so it struck me as odd to see such concern for efficiency
You have it the wrong way around, it's because of that concern that they have billions of dollars. Obsessively making things a tiny bit more efficient is their business. Works both for software and for markets
API costs are margin in trading and margin is everything. If you don’t have a way of managing that margin, you’re not a real trading firm. Pairing this type of execution with HFT would be their long term goal and that requires lots of feedback done in microseconds.
I thought the more interesting thing was that they're spending time writing custom tokenizers, regex automatons and logging frameworks at all, just so it can be in OCaml.
For the tokenizer they had three more obvious options:
1. Use a JavaScript lib and call it via FFI. Tokenization isn't a hard task.
2. Compile tiktoken to WASM and call it directly.
3. Stick with their Python server.
Writing a custom regex automaton in OCaml and then running it in the browser is surely the worst option. How does this make sense, outside of just arbitrarily writing more OCaml for the sake of it? The justification for not using their previous scheme is only that it's "grungy", without further explanation.
Every other project also seems to be like this. The justification for it boils down to nobody else using OCaml, so they need an intern army to reinvent all the wheels they're missing.
I mean, I'm sure the interns had fun and were paid well. But there's a post below by someone with a massive inferiority complex, stating that everyone is "objectively inferior" to the people at Jane Street. This post doesn't make me feel inferior, it makes me feel a sense of awe and wonder that Jane Street's traders are so profitable they can subsidise this kind of objectively excessive IT bloat. Even the author seems to recognize this, it says the post is by the guy who has "the dubious honor of having convinced the firm to start using OCaml".
The primary purpose of an internship at most major companies is hiring - you can screen candidates much better before extending a full time offer and the candidates can also get a better read on how much they'd enjoy working there. Even the second order effect of drumming up hype for others to want to work there is probably more important to many internship programs than whatever the interns might tangibly produce.
At least I can say for sure that people at Google aren't shy about how it takes a few months to get up to speed with internal tooling and not much is actually expected from a summer intern. But they built themselves into a brand on campuses, which became so mainstream there was a stupid movie made about how fun it is supposed to be to intern at Google.
And as far as finance goes, trading interns aren't even legally allowed to trade, so I wouldn't be surprised if the culture surrounding internship at Jane Street just leans more heavily into this for SWEs as well.
Relatedly - how do others deal with being objectively inferior, in basically every way, to the types of people that work at Jane Street? It always irks me that they end each of these articles with a recruiting pitch, as if most of us would ever be good enough to even get an interview, much less get an offer.
Inferiority is how you frame it. You don't need to be Jane Street level to do good things in computer science or life. I've found a job where I do cool things on a compiler and I like it. I don't feel inferior to them, instead I feel inspired from these articles to learn cool stuff and build things I'm proud off.
I can say though that when I was working on stuff I enjoyed less (backend ERP stuff), I felt much worse and would compare myself to others a lot more. So I think these feelings you are expressing can sometimes be a manifestation of not being very satisfied with your own life.
true, when you work on something you feel personally rewarded/challenged by, you rarely compare yourself to others, i think if you feel like your life is being wasted and you aren't living up to your potential then you start comparing yourself and feel bad.
> more than once I was approached by people who wanted me to help implement their projects. But I have always refused just because I think I'm not good enough. [...] have you ever done that?
Absolutely. I'd say that 90% of the time that I've been offered consulting work, I've turned it down because I know it would require some skills -- web design, graphics, SQL, linux, ruby, C++, etc. -- which I know I don't have.
I have a reputation for being very good at what I do, and it is certainly true that there are some things I am very good at... but a large part of that is that I don't do the things which I'm not very good at.
> how do you deal with self-doubt?
If you're a generalist, there's almost certainly going to be someone else who is a better generalist than you. If you specialize, it's not hard to find a niche in which you are one of the leading experts in the world -- because the group you're being compared against is losing the 99.9999% of people who never looked at that particular niche. So I'd recommend looking for a niche; because once you're the world's leading expert on something, it's pretty hard to doubt your competence in that area.
How do you find a niche though. I can definitely go looking at jobs and find things that intrigue me, but they’re so rare, that if I waste my time trying to specialize, there will be no jobs by the time I’m “good enough”. I find it incredibly rare to see specialized work that isn’t very senior.
Most niches seem accidental, and most people I know who are highly skilled in a particular one are there solely because they were in the right place and the right, narrow time.
I've basically gotten nothing for my hustle lol (except a state school degree and working at Amazon, but I don't think that's considered "accomplished" to these types of people).
What about a large percentage of the world born into poverty and barely making a living? I'm sure a lot of people will take a state school degree and think working a job at amazon is pretty awesome or you can compare yourself to newton and feel like an absolute nobody (even if you are a great scientist in your own right). Point is, there are always people to compare yourself to to feel bad, it is never ending.
I really doubt anyone that gets an offer at Jane Street would agree with that sentiment. Nobody at elite schools agrees and they all think my class of people are lazy or pitiable failures of some kind (or at least so I assume).
I was one the interns featured in the article, and I would agree with that sentiment.
Many of my colleagues and fellow interns were skilled programmers, but very few—if any—inherently possessed something that you or I don't have. The intern class was primarily composed of individuals with a curiosity for technology and circumstances that enabled their exploration.
I recognize the ease of feeling dissatisfied with one's own achievements; I doubt that receiving an offer from Jane Street for you or gaining admission into a elite school for myself would quell our insecurities. You've accomplished a lot, though I don't know how to help you recognize that.
I know a few people from Jane Street so I can easily see why one might come to that conclusion, but I also appreciate your point. Comparing yourself against others can rob you in more than one way.
Comparison is the thief of joy. Be better than who you were yesterday. Exceptional talent is also rare, don’t feel bad if you’re not exceptional. Billions will lead happy, fulfilling lives having not been.
That’s the trouble - I don’t think I’ve ever improved at anything after working at it. SATs, grades, technical abilities all pathetic after years of work, clearly unfocused and lacking because of inherent genetic defects or whatever.
“objectively inferior” is rare if it happens at all. Maybe you’re resume will never be good enough in Jane Street’s eyes to get an interview, but unless you also write fintech in OCaml, I bet you’re better than the interns at some niche in software development.
These are young interns too. If you’re a senior dev, after a few months of learning OCaml and the popular open-source Jane Street libraries, you maybecould implement any one of these projects.
I dunno, 2 of these are at top CS schools that I can't get into, and most of their intern class is from Harvard/CMU/MIT/Stanford/Waterloo and the like. I think I might actually just be objectively inferior in all regards, not just technical or in software.
I've got ~5 years of full time experience, I make decent to good money (though I don't think most people on HN would agree lol) but I'm skeptical I could build these projects.
HN often confuses $300k for low-end-of-normal for early-career US software devs in general, when it’s above most seniors and leads at most employers. You don’t hit that territory until you’re a manager-of-tech-managers, outside the smallish third hump of the trimodal dev compensation graph.
My pay’s almost 3x median household income in my medium-sized US city, but is pathetic compared to what I see thrown around regularly as examples of lowish pay on HN.
The major hft firms do pay well but new grad pay is often inflated in a few ways - signing bonus is included in “total” as if it’s recurring, first-year guarantees/targets are treated as recurring total, “this pays out over 4 years” is lumped into single-year recurring total, and you only hear about the biggest and best online.
2020-2022 high tech sector pay and huge profitability brought by the return of volatility also saw much higher base/bonus pay across the board, but the market really cooled down towards the end of 2022 longwise all the tech layoffs and HFT firm over hiring. I watched a few eye watering open offers fade away towards year end myself.
I think there’s also wildly overestimating the value of elite schools for undergrad - in undergrad you’re so far from the “state of the art” that you’d probably get about the same level of education at those schools than you would at any good (for example) European public university/US state university with a decent CS department - of which there are many and they're neither as expensive nor as selective.
Elite schools only really matter for their reputation and for research (which is done at the grad level, not in undergrad).
Still, having such low self-esteem after individually accomplishing quite a lot more than what 95% of people on Earth (not counting in history) has been able to is a sign you need to look into with some professional help.
It's not healthy, your sense of self-value seems to be completely tied to external perceptions and expectations. Answer to yourself: who are you, actually?
I’ve gotten help, none of it has changed anything. The gist of it is just reinforcement that I’m stuck being a failure, clearly unable to progress in life due to a lack of something. Probably a mix of grit and intelligence, heavily weighted on intelligence.
>Relatedly - how do others deal with being objectively inferior, in basically every way, to the types of people that work at Jane Street?
I don't have to deal with this because I don't have a weird inferiority complex towards people who work at some trading firm. The people I know who work there are decent programmers, but as luck would have it I know good programmers outside of Jane Street as well.
I don't know any good developers in Jane Street because I don't know any Jane Street engineers. This is likely because I didn't go to a top school, because I am objectively inferior.
I know many people good at their craft. But I'm nowhere close to these interns in terms of ability, compensation or respect, and my life feels basically over at this point.
One way to look at is that the more people who can figure out how to make money in a crisis without adding to said crisis, the less severe the crisis becomes. If someone can price distressed corporate bonds better, they will bid higher, and bagholders of companies that actually have decent prospects (but are hurt by market anomalies in crisis situations) can then lose less money, compared to the alternative.
As an analogy, imagine that your home is wrecked in a hurricane, but a scraps dealer is able to offer money for what's left. If there's only one scraps dealer around, they will absolutely rip you off and pay you just enough that you'd be livid and kicking but still aren't motivated to find a way to sell the scraps yourself. But if there are two scraps dealers competing for the wreckage, it'd be easier for you to get a market deal.
Scraps dealers improve their profitability by finding better ways to sell the scraps, so in a competitive environment, you could indirectly benefit from their experience. You make them sound like terrible people by saying they "make money off a disaster", but objectively their existence is a net benefit to the situation. One thing that can make this argument break down is that scraps companies
may rely a lot on information asymmetry and price gouging, but that is actually less of problem for financial transactions that can be systematically electronified, where there's a lot of regulation and said regulation is more easily enforced.
Making markets efficient is a very abstract concept that people makes people feel disconnected from the benefits. But in this abstract way, it does contribute to the economy.
Not enough smart people thinking like Jane Street (or any large trading company with a track record of managing risk), but still performing financial activities, is arguably what causes financial crises to begin with. Think people buying into schemes based on returns that are too good to be true, bundling bad mortgages without modeling it properly, or banks not taking clients' money seriously.
I agree with your example – and I believe that a lot of people's concerns are due to the fact that the way market dynamism leads to individual resilience and recovery, still necessarily involves individual instability and insecurity with root structural causes that are worth addressing.
As a contrived example: why is there a competitive market for efficient heat pumps, and a role for a "market lubricator" to profit like a logistics behemoth like Amazon, in formerly temperate climates? Climate instability and its human consequences. The good and the bad co-exist, and I'm not going to speculate any more than that.
I also agree it's kind of sad that the best minds are trying to squeeze 1bps of extra profit out of ETF trades or worse, trying to figure out why an option is implying SP500 should be trading half tenth of a cent higher. And I believe it's a deep structural issue with the misalignment of incentives of society, where individuals are not to be blamed but everyone is complicit.
Admittedly, the only thing I know about Jane Street is that the FTX sociopaths cut their teeth there, but I'm sure the rest of them are great.
You're right that market bending is too abstract for most of us to appreciate the benefits, but markets abstract away the humans, too. I don't pretend that the industry ever knows when to stop making markets efficient and ask for regulation because they're making paperclips out of all of our savings. Whatever benefit they produce should be a tool wielded by boring technocrats.
Seriously. They're in the business of finding ever more ingenious ways of shifting money from one party's pockets to another's. I can think of some important ways in which they are definitely not better than me despite their vastly more impressive technical skill.
The people that attempt and fail their interviews settle for FAANG and are probably pretty happy. Most software engineers I've met don't even know about Jane Street.
Also, definitely not "in basically every way," but very specifically logical/mathematical intelligence.
> and there's plenty of ways you're probably "superior" to me.
I guarantee you there is probably none, lol. I can't play an instrument, I can't lift, I can't run a marathon, haven't written a book, don't know a Presidential candidate - those are all normal things for the Jane Street that came from HYPSM or even the Google class of people, but extremely out of reach for people like me.
I know plenty of Jane Street (and other “prestigious” quant firm people). They’re generally smart, admittedly tend towards overachievers, and are otherwise relatively normal people. They’re not a class of god tier humans sitting above everyone else.
You are a bit delusional about what "google class of people" is, not even mentioning the delusions about the whole concept of those "classes of people" really existing on such a level.
Out of bajillion googlers I personally know well outside of work, I know exactly one that knows a presidential candidate, and it has nothing to do with the school they went to (we both went to the same public college in the south). That's just one out of tons.
> I can't play an instrument, I can't lift, I can't run a marathon
No one is just naturally "good" or "talented" at those things. Those are all just skills that no one is good at (at first), and then they practice those, and then they get better. And yes, not everyone can be Michael Jordan, but I think the bar for "I am pretty good at basketball" is way below the MJ level. That would go for pretty much all of those things you listed. It is all a question of what you are personally interested in and why. Because if you hate playing an instrument just due to its nature, then you probably would find it really difficult to put the time in to get good at it.
> You are a bit delusional about what "google class of people" is, not even mentioning the delusions about the whole concept of those "classes of people" really existing on such a level
How so? After several years at Amazon and several years of knowing people at other companies it’s apparent the ones at Amazon have a totally different standard of living - one without having Olympic medals (apparently Google publishes a list of Googler-olympians) or famous friends or ski trips. And they don’t even know people that do these things, even though it’s common at Google.
I didn’t even ever meet a Harvard grad when I was at Amazon - it’s hard to not meet one at Google or Jane Street (more so the later of course).
To remind you, I don’t know any and nobody I know knows anybody that knows a presidential candidate, but I go on Twitter for 5 minutes and see a Yale grad with 15 mutual with Vivek. It’s obvious there’s a big club of elites with accomplishments I can’t even dream of and I’m not in it or anywhere close to it.
And yeah, you gotta get good for all of those but everyone starts with some talents (and I have none). Ive been “lifting” for years with zero improvement (probably a protein thing but I’m also fat and vegetarian, the former is probably genetics too despite my failed attempts at starvation). I’ve been leetcoding for years with zero improvement. No matter what I do I’m still mediocre, still a failure no matter how much I work while everyone else gets ahead. What’s even the point of working hard if I’ve never gotten anything for it?
Nobody has ever been able to elucidate ways that I’ve ever accomplished anything, ever.
> I didn’t even ever meet a Harvard grad when I was at Amazon - it’s hard to not meet one at Google or Jane Street (more so the later of course).
I met two at Microsoft and none at Google. It mattered exactly zero to anyone. Do you, like, just ask everyone at work which college they went to? Because I don't know that info for most of my coworkers, and neither do I care. Pretty certain that most people I work with don't know where I went to school either. The only times I even remember it being brought up front and forward was when we hosted interns, because duh, they go to school somewhere and they mention it.
> one without having Olympic medals (apparently Google publishes a list of Googler-olympians)
I personally don't know a single googler-olympian or know of one. I am sure they exist, as well as that list you are talking about, but this is literally something that no one cares about. 99.99% (probably even more) of googlers are not olympians and probably dont even know what that actually means.
> famous friends or ski trips
Famous friends - you assume they made those "famous friends" from working at a tech company as a regular engineer? And what even counts as "famous"? As for ski trips, my friends working across a bunch of companies (including amazon and noname companies) do it all the time. They just don't see the need to expose it all to everyone who isn't their actual irl friend.
> I go on Twitter for 5 minutes and see a Yale grad with 15 mutual with Vivek
Bro. Having twitter mutuals is not the same as being friends with someone. Also, I am pretty sure that bragging about being friends with Vivek (if you were one) is how one gets ostracised in a lot of circles. For a lot of people in tech, admitting to supporting him would be a social semi-suicide.
> Ive been “lifting” for years with zero improvement [...]. I’ve been leetcoding for years with zero improvement.
Get a personal trainer for at least a little bit, so that they can pinpoint what you might be doing wrong and teach you proper form and advise on nutrition. Something is going wrong somewhere, unless you suffer from one of the extremely rare diseases (e.g., some thyroid ones) or you figured out how to break the laws of physics or defy all that is currently known about human body.
Get a tutor or someone to study with using leetcode. Just throwing time at "practicing" things is a waste of time. Practice has to be mindful, and focused on improving specific aspects. If you practice a piano piece, you don't just go from start to finish over and over. You break it into chunks, focus on the problematic ones, figure out why they are problematic, train yourself in those areas, get better, then glue it all together. Just running a piece from start to finish over and over is a waste of time. Having a teacher, at least at first, would net you massive benefits. It is a night and day in terms of trying to get all the basics right on your own vs. with someone who can save you the time and anquish by pinpointing how and where you can improve.
> What’s even the point of working hard if I’ve never gotten anything for it?
Mutuals with Vivek on Facebook, on his private account. Very different situation. It means you’re friends with someone worth $600 million that was making $1 million a year when he was 26 at a hedge fund nobody’s heard of.
I’ve considered a trainer but it doesn’t seem like it would be terribly effective if starving myself hasn’t worked. It really does kind of feel like I’m just a defective human if I’m so bad at everything.
> how do others deal with being objectively inferior, in basically every way, to the types of people that work at Jane Street?
How do I deal with being inferior to Alexander the Great? You just acknowledge that you are you, with your own quirks and limitations. Being surrounded by people who appreciate you makes you not really care about comparisons.
There is always someone with something, for example talent or opportunity, that you don’t have or can’t have. I guess I was lucky to come to terms with “life’s no fair” at a young age.
That doesn’t mean I’m defeatist, I’m a very driven person, but it does allow me to be more realistic without becoming demotivated.
I'm commenting because I could have (and, in fact, have) said this. There's a hole in your heart that is preventing you from accurately self-evaluating. You are not seeing the whole picture because you're in the middle of it. Go see a good therapist. Consider ketamine infusions if you have the money.
I can promise you this: you'll never hate yourself into positive self-worth, the same way you'll never fill a bucket by drilling a hole in the side.
I can recommend the book "Learned Optimism" by Martin Seligman, maybe you can benefit from it. It definitely helped me work through feelings like this.
> being objectively inferior, in basically every way
Maybe one reason that you feel like there is a big issue to deal with, is the way you're framing this. In your framing, the scope of your conclusion goes quite a bit beyond what you can actually logically deduce from the available information.
If we look at the facts, we have 152 interns who had resume and interview skills that were objectively more suited to landing an internship at Jane Street than your resume and interview skills are (presumably - it doesn't sound like you've tried, so we're not actually sure). These people have probably been preparing for this for a while, and while there is a degree of objectivity to the way Jane Street looks at incoming profiles, it's along very specific axes (probability of success in their fintech firm), and it's not perfectly objective. You're massively overstating the scope of this, they don't measure superiority/inferiority along all possible axes.
Then you might counter - no, it's not only about resume and interview skills: clearly the 3 projects described on this page are beyond what I could do as an intern, and maybe even beyond what I can do after a few years of working experience. That would be a fair point - but note that these are 3 out of 152 projects, selected to impress and inspire outsiders, and you don't know how much guidance these interns got from more experienced folks while working on this either. On top of that, many of these interns may have been actively programming since they're 10, so effectively they might have double the programming experience you have - in which case their results make perfect sense, and you just need more time to polish your skills. So extrapolating what you see here to the full set of projects and drawing far reaching conclusions about the interns based on this small sample is probably not justified.
I'm sure that these Jane Street folks are generally very good at what they do and that it would be inspiring to see them work. But I think you can give yourself more credit and spend a little bit of effort focusing on things that you're good at that aren't measured in Jane Street interviews, because it's likely that on those axes you'd generally compare favorably to most of the cohort.
At one point in my life I was motivated in a special way, worked really hard, and got to be 99.9 percentile in a quantitative test. Now I’m just a decent coder and wouldn’t stand a chance at a Jane street interview. In my opinion, aptitude is necessary, but people who can pass those interviews are differentiated by a bunch of experience and focus, mental scaffolding that takes years to build. It looks like magic from outside, but it isn’t.
My point is you shouldn’t feel inferior, you’ve just been doing different things with your brain. What’s so great about trading anyway? It’s not a particularly efficient allocation of brainpower to society.
Be nice to yourself and challenge that thought of inferiority.
"Good enough for Jane Street" often means (much more than) having middle-class parents, decent upbringing, right tutor at the right time. I know people that are not "good enough" because they were never taught how to play the game right, who don't have parents polishing their resumes for them. Of course it still involves hard work and grit, but luck and the birth lottery plays a big role. Privilege wins in this world.
Besides, do well in big tech, use that as a stepping stone to trading firms is quite a viable path I heard.
> Relatedly - how do others deal with being objectively inferior, in basically every way, to the types of people that work at Jane Street?
Would you not rather be whoever founded Jane Street and gets the largest profit share? Why aspire to be an expensive perfect instrument in someone else's hands?
Not everyone associates self-worth to monetary or financial reasons, it takes a quite quirky human (and culture) to give that much of your identity to money.
Nah I just pretend to pity them for working at a soul sucking job when with their 'superior' minds they could have been solving some real problems humanity faces.
Hah, someone a while back showed me a Jane Street job that was apparently writing/maintaining a ton of VBA. Really curious what the interview process for that job looked like.
I had years of germanic linguistics in undergrad and grad school, and i think this is an alternate ("archaic") past tense of work, so it's literally "worked", just like you said a blacksmith would do.
i mention the germanic classes because i can't at the moment think of other examples where (outside arabic etc) there's a commutativity between a consonant and vowel (work & wrokt) but i'd swear i would think of some if not trying.
It is also originally a quote from the King James Bible. I'm sure your Wikipedia page mentions that but it's just too interesting not to mention explicitly here as well :)
Worth noting: Jane Street has one of the highest paying internship programs ($120/hr) along with other trading firms [1]. Their internships are extremely competitive.
But to be honest, what surprised me most about that was that Jane Street cares about counting the token usage in its API calls. I can see why they would care if they were interacting with AI models at very large scales, but are they? They have billions of dollars, so it struck me as odd to see such concern for efficiency in what must be a very small cost center.
I wonder if the motivation is more about providing feedback to the users of the frontend tool, as they write their prompts, so they can gain an intuition for what makes an efficiently tokenized prompt? (So that when the day comes that they are running these prompts at scale, the token sizes really matter). Or maybe I'm imagining "prompts" wrong, and these are not just a few paragraphs of text, but actually 20,000+ tokens at a time, including data like e.g. pasted CSVs of trading history for the LLM to analyze?