What the interns have wrought, 2022 edition
blog.janestreet.comJane Street interns get paid very well. It's one of the few companies that probably actually compensate interns at the level of their output. These seem like cool projects. Good for the interns.
No offense but just out of curiosity have you ever had interns working under you in the past?
I havnt managed them directly but I've worked closely with two. One who was pretty much the friend of the owner and pretty useless. One who was really incredible and got a return offer right away and finished a full project while there.
Obviously not all interns are equal. I guess I should have clarified that I meant at companies like Jane Street which have strong internship programs. They have difficult interview processes, pay very well, and interns actually do interesting work.
The high level of compensation for these specific interns reflects their high level of talent.
It's extremely difficult to get an internship at a company like this. These aren't run of the mill interns who are coasting through college and trying to check the internship box.
Is anyone else blown away by the level of talent in this younger cohort nowadays? We have interns and juniors doing solid mid level work.
Their interns are easily the top 1% of CS students in the US. Some of their interns who attend CMU already have OCaml experience, as CMU uses OCaml for their compilers course. Some of their interns have interned at other trading firms.
The OP stuff is brilliant (I guess) but it's small libraries. It's not "senior" work that manages a lot of complexity and deployment risk and $MM cost/benefit tradeoffs, it's hard/smart small work. Exactly what you want smart juniors to do.
Many seniors would never be able to do the stuff in the OP.
That's definitely senior-level work
They have the only podcast worth a shit too.
bill simmons podcast is very good too
Fun fact: Jane Street interns get paid $96 / hr. They're one of the top paying internships! https://www.levels.fyi/internships/
That is high pay - you interrupted my expected post of saying other companies pay well. Most big companies (faangs but not just them) I've worked in have paid interns pro-rated wages slightly below a minimum starting pay, so maybe the equivalent of 75k. But 96/hr is 200k annually. I agree that's very high pay for an intern. Do they pay beginning devs that much? And the next step is how much for a regular experienced dev.
New grads start at ~$388k/yr. Every grad gets the same baseline offer.
Jane Street discriminates based on what school you went to, during their hiring practices.
Software development is an outlier in that companies don't tend to care where you went to school. Lots of firms on Wall Street will only hire from a few schools, or sometimes even a single school. It makes sense Jane Street would be somewhere in between.
The rub is that their discrimination practices seem to care more about prestige over quality of CS programs.
Sure, go ahead and use quality of school as a filter, but don't go around discriminating against top public schools while accommodating what are honestly fairly mediocre CS programs at certain Ivies / prestige-private-schools.
They are not discriminating by quality of CS program. They are using exclusivity of a school as a proxy for some combination of potential and pedigree.
“Pedigree”
They don't discriminate against good public CS programs though? There are many interns from Berkeley for example. There also aren't that many SWE interns from e.g. Harvard - there are a lot of trading interns from HPY but that program recruits heavily from math departments, not just CS.
I'm not sure if people ITT are mixing up trading with engineering or projecting old Wall Street biases onto Jane Street, but it's just not true that they prioritize what school a SWE went to. I'm not claiming they pour recruiting resources into Mediocre U, but they don't automatically filter out any schools, and there aren't any schools that get an automatic interview either.
And most relevant here, any school bias that does exist is related much more to the CS department at that school than any kind of archaic prestige/exclusivity factor. Waterloo and CMU are among the most represented for SWEs there - very good schools but not exactly global brands.
When I graduated I interviewed with Jane Street and I graduated from essentially a no-name state school (Suny Oswego). I think I got a fair shake out of it.
Hardly no-name: that's where Doug Lea teaches! I've been seeing Oswego on his papers forever.
You're an exception. I also went to a SUNY school and I couldn't even tell you where Oswego was.
dl is the exception I think.
This seems fine. There are some schools where it is simply not worth allocating your finite amount of recruiting resources. It by no way means that there are not super smart people at those schools, just that the percentage of students that fit your hiring criteria is too low to justify spending time at those institutions.
it's probably more like if your school isn't an ivy, stanford, mit, whatever you're resume goes right in the trash. also that logic is super self perpetuating and I imagine could be used to justify other worse forms of discrimination (not saying at Jane Street but in general)
False. When I was in college, about 8 years ago, Jane Street not only hired from our school, they came to our career fair on regular. I interviewed with them myself at the time for an internship, but fumbled because my combinatorics knowledge was severely lacking at that time, and algo wasn't quite up to par either.
My college was a public state school (Georgia Tech). While it is a great school, it was public and affordable (could be done for nearly free in-state). But it definitely was not some super difficult to get into exclusive ivy club or the likes.
So it seems like they indeer attend a rather wide variety of schools, given the size of their company. But they aren't a mega giant like Microsoft/Amazon/etc. that can afford to directly send people to recruit from almost every school in the US, their headcount was under 1k total iirc, and significantly.
GT is a top-5 engineering school in the US across many different disciplines, it's not just any public state school. A lot of places recruit heavily from them that would be less interested in other state schools (including most other state schools in GA).
EDIT:
Some context for people unfamiliar with GA: GA Tech, historically, was the public engineering university for the state. For other public universities with engineering programs, most of them were distance versions of GT with students transferring to GT for the last couple years into the 00s (changed by the mid-00s with some students finishing at the host university but graduating with a degree from GT). UGA has only had a College of Engineering since 2012. But prior to that, since the 1930s, UGA had no major engineering program outside of agricultural and a couple other things, with the primary focus of engineering education in the state being at GT.
Other schools had other focuses, but not engineering until very recently.
C’mon your counter example is that they recruit from Georgia Tech. How much recruiting does Jane Street do at UGA? George State? Spelman?
There's about four thousand four-year-degree-granting institutions in this country. How much recruiting should Jane Street do at all of them?
When I was college shopping, Georgia Tech was near the top of the list for CS. I think it was actually at the top for robotics.
sorry my list should've included some public schools georgia tech or cal berkeley. my point they would be more preferential towards top tech schools
Definitely not the case. I used to work there - lots of colleagues from outside the ivies etc. Lots of people who didn't go to college for CS, and in some cases did not go to a traditional college at all. Are there a lot of people from e.g. MIT? Yeah, because MIT is a good school and produces lots of people you want to hire.
I think this is more in regards to university recruiting, especially with internships? I have no doubt about your experience with post grad hiring, college name doesn't matter as much after you have experience under your belt
I think in general smaller companies have to prioritize on-campus recruiting, but if you sent in your resume and you had some random no-name school (or a school that specialized in some completely unrelated industries), I doubt that would disadvantage you beyond the fact that it doesn't necessarily communicate much information to hiring staff who aren't familiar with it.
That's fine. Considering the business Jane Street engages in is highly competitive (maybe even the most competitive in the world), if their hiring practices cause them to have lower productivity, the market will force them to correct their practices or go out of business.
In fact, historical evidence shows amply that the more competitive an industry, the less discrimination (of the sort that we talk about now, like on racial or ethnic grounds) is to be found; and the most regulated, monopolistic, or anti-competitive the field, the more such discrimination is found.
Independent of Jane Street, this is absolutely not true of investment or finance as a whole: there are dozens of large firms where entire management layers are made up of the corporate equivalent of "legacy" admissions.
Finance isn't a purely deductive field: you can't purely compete and reason your way past dumb luck and overwhelming access to liquidity. Institutional players have both, and are more than happy to hire their top performers' mediocre sons and nephews to keep them happy.
A classic rebuttal : https://danluu.com/tech-discrimination/
Thank you for pointing me to this -- apparently it's not so classic that I have read it already. ;-)
Since I assume you didn't write it, I won't go into a point-by-point rebuttal of the points raised in this piece, but in the interest of keeping up the debate I will mention a few things.
Like many who think along these lines, there is a base fallacy assumed here by the author, which is that absent discrimination, the distribution of tech workers will resemble the distribution of the larger population. Perhaps it is the classical liberal tradition (that "all men are created equal") that leads us to believe this, but there is no reason to believe that any slice of the population must reflect the broader whole, no matter the subject at hand. Over half of the players in the NFL are black and only 0.1% Asian, out of about 13% and 6% of the general population respectively. About four-fifths of all cab drivers in the US are male. Billions of people enjoy running around the world and many millions compete in it, but the Kalenjin of Kenya comprise a stupid number of those at the very top levels of distance running. It's entirely plausible that Chinese, Indians, and Vietnamese (as mentioned) enter the profession at higher rates and succeed at higher rates than other groups without involving a cabal that favors them over other groups, or indeed even personal biases for or against.
This isn't to say that personal biases don't exist and the numerous anecdotes of objectionable behavior encountered by individuals who don't fit the typical "tech bro" mold are invalid. But words like "discrimination" must be precisely defined, and I favor one where it means legal or policy-based exclusion or subjugation of certain groups, contrasted against personal biases ("racism" when it comes to ethnic groups, "sexism" in gender, etc.). When older folks get fewer callbacks from interviews or women are seen as girlfriends at tech conferences, that's personal biases at play, but it is not the same thing as being discriminated against. When you conflate the two you head down a dangerous path that we're treading as a society now, which is the tendency towards totalitarian control of people's thoughts; and this cost must be weighed against any benefit.
It's interesting that the author specifically mentions the Townsend-Greenspan firm vis-a-vis the gender pay gap among economists, because that was a very example of where capitalists motivated by profit took advantage of the fact that women were paid less for equal work, hired such women for slightly more than their competition while still getting the same output, thereby both raising women economists' wages and making out with a handsome profit. It's an argument for letting the market play out. (I do question why the firm dissolved when Greenspan was appointed Fed chairman, instead of continuing on with their more competitive labor.)
This is getting long, so I'll conclude by reiterating what I see as two very salient points from the essay: "The market is just humans. It's humans all the way down", and "We can fix this, if we stop assuming the market will fix it for us".
Yes, the market is humans all the way down. The beauty of free markets is that it leaves the decision-making to the individual humans involved, and not an enlightened group of elites who think they can fix it by meddling. History is littered with examples of people who thought they can "fix this" and ended up making it worse -- so much so that we have a saying about it.
If you're interested in exploring this angle more, I'd be happy to point you to a few works in the literature who have made the argument way more clearly than I ever could.
I highly doubt a multibillion company engaging in more widespread recruiting practices would be the difference between them succeeding and failing
I think you grossly overestimate the percentage of candidates at lower tier schools who would be qualified to pass a Jane Street (or really any HFT firm) interview. Even at the top schools the pass rate is incredibly low. In expectation you would be getting probably <5% of your new graduate/intern class from the entire bottom 80% of schools.
The quality level drops off very sharply after you go below the top 20-30 CS schools. It's just a matter of limited resources for recruiting. The common path taken by people who go to "lesser" schools is to grind really hard to get into a good MS or PhD program and then go to these firms.
to be upfront, I don't buy that a company with such large resources couldn't recruit from a larger pool of school. I'm also not under any illusion about the caliber of student at upper vs lower ranked schools, I see it as a matter of equity and meritocracy
Why not? Having a company with multiple billions of dollars doesn't in itself produce any returns without high quality labor to work it, especially in competition against other such multi-billion dollar companies. It might not fail this quarter or next, but inevitably they'll be chased out, if in fact the recruiting practices lead to lower productivity.
I got an internship offer from Jane Street back when I was at NYU, so at the very least the criteria are (or used to be) broader.
Why exactly is that "discrimination" and not, you know, simple evaluation? Is every company that doesn't hire based on a random number generator discriminating?
if not in <list of blessed schools>: skip_interview_entirely return False
...as opposed to give every person a fair chance based on merit.
That's not Jane Street's process at all. I know first hand people from schools you've never even heard of that got internship offers, and I also know first hand people from MIT that got resume screened out. I'm not going to claim that there is no school bias at all, I don't know the finer details - but it's definitely not explicit as you describe. There are a lot of comments all over the internet (including this thread) from people at non-target schools that were interviewed.
There are some trading companies that do seem to legitimately do that kind of filtering (Two Sigma's application has a drop down menu for school with like 10 options and then "Other", DE Shaw asks for your SAT score, etc.). So I can see where the idea came from, but it's not true of JS.
In the trading industry, DE Shaw and Five Rings are the two places that care about school prestige well into someone's career. All other firms don't give a crap about it once someone has some experience.
I went to a crap school in India and a mediocre US school for my grad work, still got interviews at trading firms as a new grad and an industry hire. Currently working at one right now, so I can say that I have seen it first-hand.
I have proof otherwise that can probably be replicated if need be. I can DM you the details. Granted this was a few years ago so hopefully they've matured since then.
Real life has to deal with things like time and energy constraints.
Many (most?) companies do. They make the tradeoff that the false negatives (those from non-targets who would be hired) are worth losing out on in favour of the smaller amount of applications to process because among that smaller pool they have a higher true positive (those from targets who do get hired). In stats/ML speak, they favour high precision [0].
Why? That seems like a dumb way cutting of potentially very talented candidates. Seriously it removes over 90% of candidates, if not more. That seems unsound.
You know there's a very well known joke that it's hard to get fired as a trader for underperforming the market by buying Apple, well it's hard to get blamed for hiring a bad intern if you hire him from MIT.
> well it's hard to get blamed for hiring a bad intern if you hire him from MIT.
Whew! I had four paid internships while I was at MIT and for two of them I got nothing meaningful accomplished. I learned a lot from that but for those two I was of no use to the companies.
Even decades later I still feel bad about that, so it's nice to think that nobody suffered as a result.
The goal is to hire efficiently, not hire every single qualified candidate. If 10% of MIT graduates are qualified but only 1% of graduates from a no-name school are qualified, why spend 10x the effort to try to recruit from the latter?
Given that they may only hire a handful of interns each year, then that actually makes sense.
Jane Street gets a shit ton of applicants. Every slightly ambitious college grad applies there. Same for experienced roles, it's full of FAANG folks hoping to get in.
Their aim is not to hire every good candidate. Maybe they are fine with just 10% of the candidates if they could be top choice of good enough chunk of candidates.
Dunno about this, I did my MSc at McGill and got interviewed by Jane Street (twice).
That's a really good uni, no?
My understanding is that it's basically the Canadian Harvard. Is the reputation worse than I thought?
There’s no such differentiation in Canadian universities. There’s a few that are a bit better than others in some areas.
No, that's Waterloo.
I thought that was the Canadian MIT (heavily focused on CS)?
Jane Street likely has the highest applicant to available role ratio of any company in the world. I'm sure some heavy filtering is required, and probably useful.
That's ok, some of the universities reject applications from PO Boxes or application fees paid with money orders. The filters work from a certain point of view, but I cannot help but wonder how much talent sits on the sidelines because nobody has time or the inclination to look.
Lots!
The statement that "talent is evenly distributed across the world, but opportunity is not" is very, very true.
I work in the HFT world. They might do it for new graduates, not so much after that.
I went to a school you never heard of, in a former communist country, and Jane Street was happy to interview me for internship (I didn’t make it, sadly, but I can only blame my own lousy performance).
Pretty impressive projects, did they only have 3 interns though?
> And once again, the program was bigger than ever. We had 122 dev interns globally, coming from 42 different schools and 13 countries. As the program grows, it gets harder and harder to summarize it with any degree of fidelity by picking just three projects.
I wish this evil company would stop spamming the internet for PR.