YUZHENG SUN · 课代表立正
Some work ends when it ships. Some keeps changing what others can do—and keeps representing the person who made it. I am building Superlinear Academy to become the latter.
Cornell Economics PhD · Former Amazon, Meta, Tencent & Statsig · Founder and author
WHAT I BELIEVE
Many capable people learn, rise, manage, and deliver. Then the task resets, the title moves on, and it becomes hard to say what their best attention added to the world. The deeper desire is to make something that could only have taken this shape through their own way of seeing and working.
AI is making the first version cheap, but it will not make a defining work for you. New leverage should not lower our standard; it should let one person or a small team finish what used to be entirely beyond reach.
THE HORIZON
The products of Steve Jobs and Zhang Xiaolong. Lionel Messi's football. Christopher Nolan's films. Geoffrey Hinton's and Warren Buffett's convictions surviving decades of disagreement. Tsunekazu Nishioka's craft, giving ancient timber another thousand years of life. Different fields and temperaments; the same unmistakable result: the work carries the maker.
WHAT LASTS
People shape work, and work shapes people. A defining work is not something you think right once; through version after version, materials, masters, users, markets, and time answer back until it slowly becomes real.
THE STORY
I learned how to succeed inside schools and companies. The harder change was to put my best attention behind problems, standards, and work I had chosen—and accept responsibility for what happened next.
CORNELL · AMAZON · META
Economics trained me to ask about causes, choices, incentives, and constraints. Product work made those explanations answer to a harder question: did they change a real decision for real users?
TENCENT · 30-PERSON DATA & AI TEAM
Leading a team taught me that one person seeing clearly was not enough. Priorities, incentives, standards, and ownership had to hold together before an idea could change an organization.
2022 · A DIFFERENT BET
After leading Tencent's 30-person Data & AI team and repeatedly earning its highest performance rating, I left a clear management path, returned to the U.S. as an individual contributor, and began turning work I had built outside my day job into something I was willing to carry for the long term.
STATSIG · SUPERLINEAR
At Statsig, experiments made belief revision part of the product workflow. At Superlinear, courses, community, books, tools, and public work remain open to the same question: are they genuinely useful after the first release?
JUDGMENT ON THE RECORD
APRIL 2021
Define the limits of the dominant AI paradigm
FEBRUARY 2023 · BEFORE GPT-4
Explain what changed—and reason forward
Two public videos in 2021 located the earlier model's boundary in its mechanism: it learned mappings from labeled data and automated narrow settings well, but lacked general understanding. ChatGPT's instruction following, in-context learning, and apparent reasoning changed that model. From there I reasoned forward to falling inference costs, tool-connected workflows, persistent memory, direct result delivery, and ChatGPT-native systems. The original probabilities remain on the page; so do the two major errors added later.
Read the original argument and later reviewSUPERLINEAR ACADEMY
Not merely a course, and not merely a community. I want people with professional depth to come here and turn their judgment into products, work, and ventures. Superlinear is the most important long-term work I have chosen—and it is still far from finished.
The free community, AI Builders, the membership, and our enterprise programs each solve a different part of that problem. The pages below show what each one is responsible for—and what already exists.
OPEN, PUBLIC VALUE
A community of 20,000+ people seriously using AI. Unfinished work can meet deep analysis, real projects, first-hand experience, complementary people, collaboration, and opportunity.
20,000+ members · 600+ public project posts · nearly 7,000 project comments
Join for freeSTRUCTURED LEARNING
A structured learning system I teach with Yage (Wang Yan), a Columbia electrical engineering PhD and AI researcher with nearly 40 papers, including work at leading AI conferences such as CVPR, NeurIPS, and KDD. The aim is to understand AI deeply enough to make reliable choices and build beyond a tutorial.
3,000+ paying learners · 5.0/5 on Maven
Explore AI BuildersYEAR-ROUND MEMBERSHIP
A year-round source of deep analysis, guest masterclasses, monthly Q&A, core courses, Skills, recordings, and practitioner context—so the next important decision can begin from a higher place.
12+ masterclasses · monthly Q&A · 3 core courses · selected Skills
Explore the membershipORGANIZATIONAL PRACTICE
We work with teams to bring new AI capability into real workflows, evaluation, ownership, and organizational constraints—so an experiment can become a way of working the organization can trust.
Tencent · Xiaohongshu · Meituan · DoorDash
Explore enterprise work“Yuzheng distills years of product growth wisdom into actionable insight—helping data scientists surface decisive signals, PMs turn numbers into strategy, and founders find a repeatable path to compounding PMF.”
200+ CONVERSATIONS
I talk with them about technology, products, and companies, but also about the moments that changed a consequential choice: what they saw, why they were willing to bet, and how reality rewrote the answer. These conversations keep raising my own standard for good work.
The work reaches a 400K+ audience across YouTube, Bilibili, and Xiaohongshu, including engineers, researchers, founders, investors, and operators across China and the U.S.
BOOKS
Growth Data Analytics Playbook covers product-market fit, metrics, and experimentation. 真本事 asks how work becomes capability, leverage, and income of your own. Different subjects, same attempt: preserve judgment in a form another person can use.
Explore both books
FROM PEOPLE WHO KNOW THE WORK
“Yuzheng has built an AI education community that is carefully curated, practical, and unusually useful for people doing real work.”
“Yuzheng shares more than tools. He teaches a philosophy of thinking—the rarest and most valuable thing in the AI era. Even as a scientist, I found his course deeply illuminating.”
YOUR WORK
You do not need a complete answer before you begin. Come see what people here are making, or bring an unfinished question and put it in contact with knowledge, peers, and reality.
