When electronic trading is better than an AI job at Meta
If you're a computer science student with a focus on artificial intelligence and natural language processing (NLP), then working for Meta's NLP teams in Menlo Park might seem like the manifestation of your dreams. Andrew Mao thought so, before discovering a latent passion for market microstructure. Now he works as a quant in electronic trading instead.
"At Meta I worked on an NLP research team," says Mao. "I was building some tools to help develop language models for things like summarizing meetings. We were taking texts and transcripts from conversations and using AI to generate concise summaries and answer questions."
Mao studied computer science at Maryland University and focused on AI. His moment at Meta was in 2021, when he spent three months interning for the social media firm at the height of the pandemic. Instead of joining Meta full time, though, Mao took a completely different job as a quantitative software developer at electronic trading firm DRW in Chicago after graduation. He knew what he was getting into, having also interned at DRW the previous year. His job at DRW doesn't have anything to do NLP, or even with AI, but Mao says that's fine: his current role is more stimulating than his AI work at Meta.
"I work on strategic systems that provide the brains for the trading systems," he says of his work at DRW. "It's about the quantitative logic based on the risk and pricing calculations behind the trades. Most of my work is done in Python."
To do his job, Mao says you need to know "quite a bit about trading." You also need an enthusiasm for market microstructure, which Mao describes as, "the study and behavior of markets when you zoom in closely at very short time scales," he says. Those time scales are milliseconds (a thousandth of a second) and below. "They're well below human reaction time," says Mao. "We look at everything that can be observed in the market when it comes to the interactions between different orders on a localized basis, and to large scale macroeconomic events. There's a lot of mathematical theory behind risk calculations and options pricing."
DRW has over 100 traders in Chicago; Mao and his quant engineering colleagues work alongside them. "I've spent time sitting next to them on the floor and discussing their needs," he says. "The traders are the end users."
This is where working in electronic trading at DRW also differs considerably to working in AI at Meta. At Meta, Mao says much of the work during his internship was research-focused, designed for nebulous end users, and was never deployed to production. "Here, there are much tighter feedback loops," says Mao. "We deliver things to the end users and can see the outcome. It's driven by the markets and is very fast-paced."
Before settling on electronic trading, Mao also spent time interning in the defense industry, but he says this was equally unfulfilling. "Most of the interesting work in defense requires security clearance and as most of us interns didn't have that, we didn't get to work on any of the interesting projects," he says. "Instead, we were siloed into pet projects with little chance of making it into the real world."
When he's not building electronic pricing models, Mao is climbing mountains. "I've climbed at both Devil's Lake and Governor Dodge in Wisconsin," he says. "Climbing is quite similar to the work that I do as an engineer - the different route up a mountain are a kinetic version of problem-solving. This is probably why there are a lot of engineers in the climbing community."
Have a confidential story, tip, or comment you’d like to share? Contact: +44 7537 182250 (SMS, Whatsapp or voicemail). Telegram: @SarahButcher. Click here to fill in our anonymous form, or email firstname.lastname@example.org. Signal also available.
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 libelous (in which case it won’t.)