OpenAI has long admired financial services quants and financial services quants have long admired OpenAI. It helps that people leaving hedge funds and electronic trading firms can go and work there while serving the lengthy non-competes typical of the finance industry.

Now it seems that Open AI is stepping up its financial services hiring efforts. Open AI’s CEO Sam Altman tweeted an open call to HFT developers in april. Business Insider reported this week that OpenAI is now offering $3m for quants with just a few years’ experience. 

That might be tempting. But do you know what you’re getting yourself into?

Get Morning Coffee ☕ in your inbox. Sign up here.

Calvin French-Owen, a recently departed MTC (member of technical staff) at OpenAI  posted an extensive review of his experience at the firm earlier this week. In it, French-Owen – who was described as “one of the best” by ChatGPT’s head of engineering, revealed seven things that might surprise you about the OpenAI culture:

1. There is no ‘master plan.’

Given the cult of personality surrounding Sam Altman and his quest for artificial general intelligence (AGI). you might think OpenAI is ruled by Altman’s strategy. Not at all, says French-Owen. OpenAI didn’t have a roadmap for the next quarter at the time he joined.

Instead, French-Owen said that teams naturally form around which research ideas most interest them, and that “most research gets done by nerd-sniping a researcher into a particular problem.” Boring ideas, or problems that don’t demand immediate attention, are disregarded. As good ideas emerge, French-Owen said OpenAI is able to change direction “on a dime,” but when it commits to something, it “goes all in.”

2. OpenAI is a Python house, but has fans of Rust

Regarding OpenAI’s tech stack, French-Owen notes that the firm has a “giant monorepo” built mostly in Python. However, it also has “a growing set of Rust services and a handful of Golang services.” He said the abundance of Python creates a “strange-looking codebase because there are so many ways you can write Python.”

Because of that amorphousness, you’re expected to code with “a lot more guardrails,” making sure that code works by default and is hard to misuse.

3. You’ll work until 4am some nights

Employees of OpenAI and other AI firms have told us that you’ll end up working much more hours in AI research than you were in finance. 

French-Owen confirms this. In a sprint preparing to launch OpenAI’s coding agent, Codex, he said he was regularly working 15-16 hour days. The night before launching, he and four others “stayed up until 4am trying to deploy the main monolith.”

OpenAI staff seem to enjoy that grind. One MTC we previously spoke to said that work-life balance doesn’t exist there, because “nobody wants a life outside of deep learning.” Regarding Codex, French-Owen said “I’m not sure I’ve ever worked on something so impactful in my life.”

4. The team you join matters.

Not all teams are working at that pace; French-Owen said they can “vary significantly in culture.” He said some teams move at a more consistent pace while others are “sprinting flat-out all the time.”

He also notes that experiences differ greatly depending on whether you work in research, applied or go-to-market teams. The MTC we spoke to previously told us that there’s a lot of snobbery among long-serving members of research teams, who view themselves as superior to applied colleagues. Given the rate at which OpenAI is growing, its unlikely you’ll encounter too many of them, however.

5. There will be internal competition

Internal competition will be nothing new if you’re a quant from a pod-based hedge fund, but quants from friendly firms like Hudson River Trading might want to take heed. French-Owen said there’s a “strong bias to action” and that successful researchers build products “without asking permission.” He said that there were 3-4 other teams working on coding agent prototypes at the same time he was.

This bias to action is reflected in OpenAI’s promotion structure. French-Owen said promotions are “primarily based upon their ability to have good ideas and then execute upon them.” Office politics and the ability to present in an all-hands are less important.

6. You’ll see regular Slack messages from Sam Altman

French-Owen said one of the more idiosyncratic aspects of OpenAI is that “everything runs on Slack,” so much so that he only received 10 emails during his entire tenure. 

You’ll see frequent messages from all the OpenAI executives, including Sam Altman and Greg Brockman, who French-Owen said are “quite visible and heavily involved… there are no absentee leaders.”

7. You’re swapping PnL for Pro-Subs and ‘GPU math’

French-Owen said that at Open AI “everything is measured in terms of ‘pro subs'” [professional subscribers to ChatGPT] and that you can immediately see how a new product launch has impacted that figure.

On the flipside, French-Owen said that “everything is a rounding error compared to GPU cost,” and you have to carefully calculate the load capacity required to launch a new product. Lowering latency helps, so it’s no surprise that Open AI is pursuing quants working in high-frequency trading.

Have a confidential story, tip, or comment you’d like to share? Contact: Telegram: @AlexMcMurray, Signal: @AlexMcMurrayEFC.88 Click here to fill in our anonymous form, or email editortips@efinancialcareers.com.

Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libellous (in which case it won’t.)