On Oct. 19, 2026, Meta Chief AI Officer Alexandr Wang posted a selfie on X with a seven-word caption: 「got the HRT gang back together!」 The photo looked casual. The reaction was not.
What drew attention was the story behind the people in it. The three men in the picture, along with several former classmates who were not there, were part of Hudson River Trading’s first intern class in the winter of 2015. The program took just 10 people. Ten years later, companies founded or led by that group are worth more than $100 billion in aggregate, according to the source article.
Wang appears on the right. The article describes the 29-year-old Scale AI founder as having been brought to Meta last year in a $14.3 billion deal, where he became chief AI officer and took charge of Meta’s superintelligence lab.
On the left is Jeff Yan, founder of Hyperliquid. The source says the decentralized exchange, built by an 11-person team without outside fundraising, generated $900 million in profit last year and reached a market value of $76 billion.
In the middle is Scott Wu, founder and CEO of Cognition. His AI coding agent Devin is said to be growing enterprise customers at a monthly rate of 50%, and the company’s latest valuation is listed at $48 billion.
The article says the last time all three may have been in the same room was likely that winter in 2015, inside HRT’s office by the Hudson River.
They knew each other before HRT
Their connection did not begin with the internship. Jeff Yan and Scott Wu both won gold medals at the International Olympiad in Informatics, or IOI. Wu won three golds and finished first with a perfect score in 2014. Alexandr Wang was a regular in U.S. computer science and math Olympiad circles.
That world was small enough that many of them had already crossed paths in camps and training programs while still in high school. In the winter of 2015, they converged again under one name: Hudson River Trading.
HRT is described in the article as one of the world’s top quantitative trading firms, headquartered in New York, with daily trading volume that at one point accounted for more than 15% of the U.S. stock market. That winter, the firm ran its first internship program. There were only 10 spots. Yan was a junior at Harvard, Wu was a freshman at Harvard, and Wang was a freshman at the Massachusetts Institute of Technology.
The internship lasted three weeks and was structured like a competition, with rankings after each round. Jeff Yan later recalled that the top two in every round were always him and Scott Wu.
The article cites a long-form interview with Jeff Yan in Colossus magazine, saying Yan first heard about Bitcoin during the HRT internship through recommendations from two former partners. He did not act right away, but soon found Ethereum’s yellow paper and read a description that stayed with him: a computer run by the whole world that no one could shut down. The article frames that moment as an early seed for Hyperliquid.
That same winter, a 19-year-old Wang said something at his birthday gathering, surrounded by friends from Harvard and MIT, that would later be quoted repeatedly: 「Why not us?」
Three founders, three paths
Alexandr Wang: from data labeling to Meta’s AI command center
After HRT, Wang made a decision that surprised people around him: he dropped out of MIT.
In 2016, at age 19, he co-founded Scale AI with Lucy Guo. The company focused on labeling training data for AI systems. The article argues that Wang’s key call was to skip over any single AI application and instead bet on the infrastructure layer serving the entire industry.
As AI spending accelerated, Scale AI became a core supplier to OpenAI, Meta, and the U.S. Department of Defense, according to the source. In 2021, the company’s valuation rose to $7.3 billion, making Wang, then 24, the world’s youngest self-made billionaire. In a 2024 financing round, Scale AI reached a $13.8 billion valuation, backed by Nvidia, Amazon, and Meta.
Then came the June 2025 transaction that, in the article’s telling, shook Silicon Valley. Meta bought a 49% stake in Scale AI for $14.3 billion, implying a $29 billion valuation. As a central part of the deal, Mark Zuckerberg wanted Wang himself to join Meta as chief AI officer and lead the newly formed Meta superintelligence lab.
The article presents that shift in stark terms: a 28-year-old founder became the person directing AI strategy at one of the world’s largest social media companies.
Jeff Yan: from market making to on-chain trading infrastructure
Yan took a different route. His focus became crypto market structure.
After graduating from Harvard, he did not start a company immediately. He worked at Google and at a self-driving car company, while continuing to watch crypto closely. In 2020, he founded Chameleon Trading, a crypto market maker. That experience gave him a direct view into what the article calls the structural weaknesses of centralized exchanges. The collapse of FTX then gave him the final push.
In 2022, Yan shut down the Chameleon team and began building Hyperliquid with his own capital. The project is described as a decentralized perpetual futures exchange running on blockchain rails. There was no venture capital funding and no token presale. The team had 11 people.
His product goal was straightforward: bring a quant trader’s standards for low latency, high throughput, and execution quality into an on-chain system, and get as close as possible to the user experience of centralized exchanges, or even surpass it.
The article also says Intercontinental Exchange Chief Executive Officer Jeffrey Sprecher once described Hyperliquid’s trading activity as 「greater than Nasdaq.」
By the figures cited in the source, Hyperliquid generated more than $900 million in profit last year with only 11 employees. Four years after launch, its market value had reached $76 billion.
Scott Wu: building software that writes software
Wu is presented as the most decorated contest competitor in the group. The article lists three IOI gold medals, a perfect-score championship in 2014, a national MathCounts title, and a third-place finish in Google Code Jam.
After two years at Harvard, Wu dropped out and co-founded the AI social network Lunchclub. The company that made him a bigger name in Silicon Valley was Cognition, founded in late 2023 with Steven Hao and Walden Yan. Together, the three founders held five IOI gold medals, the article says.
In March 2024, Cognition launched Devin, which it described as 「the world’s first AI software engineer.」 The product was presented as being able to plan, write, test, and deploy code on its own.
Devin sparked heavy debate in the tech world. The market response, as laid out in the article, was clear enough. Cognition’s valuation climbed from $350 million at the start of 2024 to $10.2 billion in September 2025, then to $26 billion in May 2026, when it raised $1 billion. Its annualized revenue rose from $5.4 million in mid-2024 to nearly $500 million by mid-2026.
By September 2026, annualized revenue was nearing $1 billion, and the latest valuation was reported at $48 billion. Wu also said more than 90% of Cognition’s internal code is now written by Devin itself.
The article sums that up this way: a three-year-old startup went from a $350 million valuation to $48 billion, up more than 130-fold.
The network extends well beyond the selfie
The photo shows only three people, but the article argues that the HRT 2015 class and its surrounding Olympiad circle reach much further.
Scott Wu is cited as saying the internship was not the beginning of these friendships. Many of them had already known one another through Olympiad competitions in high school, and that small circle later produced a wider set of founders and senior operators.
Jesse Zhang: from social gaming to AI customer service
Jesse Zhang was another member of the same HRT intern class. His earlier social gaming company, Lowkey, was acquired by Niantic, the developer behind Pokémon Go. In 2023, he founded Decagon, an AI customer service agent company focused on replacing enterprise support workflows with AI.
The article says Decagon reached a $4.5 billion valuation in less than three years. Its customers include Duolingo and Hertz, and Zhang was named to the Forbes 30 Under 30 list in 2026.
Johnny Ho: from quant trading and Quora to AI search
Johnny Ho, another Harvard alumnus, won three IOI gold medals. The article says he knew Scott Wu and Jeff Yan as early as high school. In a 2013 photo of the U.S. IOI team, he stood alongside Steven Hao and Scott Wu.
After leaving the contest world, Ho worked as a quant trader at Tower Research Capital and later joined Quora as an engineer. In 2022, he co-founded AI search engine Perplexity with Aravind Srinivas and others, and now serves as co-founder and chief strategy officer.
Perplexity’s latest financing reportedly valued the company at about $30 billion, according to the source.
Demi Guo: from Olympiad medalist to AI video generation
Demi Guo, identified in the article as Guo Wenjing from Hangzhou, won an IOI silver medal in 2015. She earned a mathematics bachelor’s degree and a computer science master’s degree from Harvard, and was previously the youngest full-time researcher at Meta AI Research.
In 2023, she left a Stanford PhD program and co-founded AI video generation platform Pika with classmate Chenlin Meng. The article puts Pika’s valuation at $470 million and says it has more than 16 million users. Guo also interned at HRT and Quora, placing her in the same broader orbit.
Steven Hao: from Scale AI to Cognition
Steven Hao won an IOI gold medal in 2014 and graduated from MIT’s mathematics department. He was an early core engineer at Scale AI under Wang, then later joined Wu at Cognition as co-founder and chief technology officer.
Citing Forbes estimates, the article says Hao, now 30, has a net worth of about $1.3 billion and is the youngest American on the 2026 Forbes 400 list.
His path also illustrates the internal talent flow inside this network. The article says he carried engineering experience and an understanding of AI data from Scale AI into Cognition. It also notes that Cognition’s three founders, Wu, Hao, and Walden Yan, hold five IOI gold medals between them, and ties that background to the company’s core architecture for AI coding agents.
Why quant firms became a founder pipeline
The article argues this concentration is not an accident.
Earlier this month, when Wang reposted the 2015 intern class photo, he added a line: 「HRT 2015 intern class is the new-gen PayPal Mafia.」
The comparison is deliberate. The PayPal Mafia is the well-known group that includes Elon Musk, Peter Thiel, Reid Hoffman, and David Sacks, who went on to build companies such as Tesla, Palantir, LinkedIn, and Yammer.
But the article says the HRT group differs in one important way. The PayPal Mafia was tied together by one company and one shared startup experience. The HRT group was connected earlier and more deeply, through a shared way of thinking.
That mindset, in the article’s framing, comes from two places: Olympiad training and quantitative trading. Contest competitors spend their teenage years learning to search for optimal solutions, pushing hard on efficiency, and building intuition for complex systems. At HRT, those habits were amplified. They learned how to extract signal from noise, make decisions in millisecond windows, and build systems that optimize themselves.
The article quotes Jesse Zhang as writing: 「This generation of people who grew up doing math, programming, and science competitions make up a disproportionately high share of founders today.」
It also mentions Jane Street, Citadel, DE Shaw, and Two Sigma as firms associated with similar stories. Even so, the density of the HRT 2015 class stands out. Out of 10 interns, at least four became unicorn founders, and the companies tied to the group are worth more than $100 billion combined.
The source goes a step further and says this founder conversion rate exceeds that of talent-rich tech companies such as Google and Meta. Its deeper explanation is that quant firms do more than train technical skill. They also teach an instinct for market structure. Wang applied that to AI data infrastructure. Yan applied it to on-chain financial infrastructure. Wu applied it to software engineering automation. Zhang applied it to customer service. Ho applied it to search.
Different sectors, same method: find an inefficient system, then rebuild it with algorithms and engineering.
From contest kids to the center of the tech cycle
The article closes by widening the lens beyond valuations and company names.
Ten years ago, these people would have been described as math champions, coding obsessives, and the kind of students who often faded into the background socially. The word used for them was often 「nerd,」 and for a long time it was not meant as praise.
After 2024, the article argues, the environment changed. As AI became a base layer for more industries, and as code itself began writing code, the people best positioned to shape that shift were often the ones who had spent years living inside algorithms and formulas.
Wang, Yan, Wu, Zhang, Ho, Guo, and Hao are presented as examples of that transition, moving from Olympiad stages to the center of the current technology cycle. The article ends on a simple note: the people once dismissed as nerds have entered their moment.