DeepSeek Founder Wenfeng Liang | Generated by AI
Wenfeng Liang: The Founder of DeepSeek
Wenfeng Liang (also known as Liang Wenfeng in Western naming conventions) is a 40-year-old Chinese entrepreneur and AI pioneer. Born in 1985 in the rural village of Mililing in Wuchuan city, Guangdong province, he grew up in a modest household where both parents were primary school teachers. From an early age, Liang showed exceptional talent in mathematics and technology, often reading comic books while excelling as a straight-A student. His curiosity extended to tinkering with ideas that foreshadowed his future: even as a university student, he was writing algorithms for stock selection using early AI concepts.
Education and Early Experiments
Liang attended Zhejiang University, one of China’s top engineering schools, where he earned a bachelor’s degree in electronic information engineering in 2007 and a master’s in information and communication engineering in 2010. His thesis focused on object-tracking algorithms for low-cost PTZ cameras, blending AI with practical surveillance applications—a hint of his lifelong interest in machine learning. During the 2008 global financial crisis, while still in school, Liang assembled a team of classmates to scrape financial market data and experiment with quantitative trading models powered by machine learning. Post-graduation, he relocated to a budget apartment in Chengdu to test AI across various fields, but initial ventures flopped until he zeroed in on finance.
Building a Quant Finance Empire: High-Flyer
In 2013, Liang co-founded Hangzhou Yakebi Investment Management with fellow Zhejiang alum Xu Jin, laying the groundwork for AI-driven trading. By 2015, at age 30, he launched Ningbo High-Flyer Quantitative Investment Management Partnership with two engineering classmates. High-Flyer revolutionized quantitative hedge funds in China by relying on mathematics, AI algorithms, and machine-driven decisions to predict markets and execute trades—eschewing traditional human portfolio managers. The firm grew explosively: by 2019, it managed over 10 billion yuan ($1.4 billion) in assets, and by 2021, it surpassed 100 billion yuan ($14 billion), catapulting Liang into billionaire status with a net worth estimated at over $1 billion.
Liang’s philosophy here drew inspiration from Jim Simons, the quant trading legend behind Renaissance Technologies. In a 2019 speech at China’s Private Equity Golden Bull Awards, he declared quantitative investing as “server-driven,” aimed at boosting market efficiency. He even penned the foreword for the Chinese edition of Simons’ biography, The Man Who Solved the Market, emphasizing the mantra: “There must be a way to model prices.” However, regulatory crackdowns on quant funds from 2019–2023 forced a pivot. High-Flyer wound down its main product in February 2024, but not before Liang had quietly stockpiled thousands of Nvidia GPUs starting in 2021—before U.S. export bans kicked in—for what partners dismissed as a “hobby” side project.
From Side Hustle to AI Disruptor: Founding DeepSeek
DeepSeek was born in May 2023 when Liang announced High-Flyer’s shift toward artificial general intelligence (AGI), with the company formally launching in July. Funded internally by High-Flyer’s profits (VCs balked at the long timelines), DeepSeek started with 10,000 Nvidia A100 GPUs and a lean team of about 100 researchers—mostly fresh PhDs and grads from elite schools like Tsinghua and Peking University, hired for passion over pedigree. Liang fostered a flat, bottom-up culture: no rigid hierarchies, instant access to compute resources, and cross-team collaboration. Ideas bubble up organically; one key innovation, the MLA (Multi-head Latent Attention) architecture, stemmed from a young engineer’s personal curiosity.
The company’s breakthrough came with open-source models like DeepSeek-V2 in 2024, which slashed training costs through efficient techniques like pruning, quantization, and knowledge distillation—avoiding the “arms race” of ever-larger models that burn billions. In January 2025, DeepSeek-R1, a 671-billion-parameter reasoning model trained on just 2,048 H800 GPUs for $5.6 million, rivaled OpenAI’s top offerings while costing a fraction. Its app rocketed to #1 on the U.S. iOS App Store, outpacing ChatGPT downloads globally. This triggered market mayhem: Nvidia shares plunged 17%, erasing over $1 trillion from U.S. stocks as investors questioned the viability of Western AI giants’ massive budgets.
Liang’s hometown hailed him as a “pride” during 2025 Lunar New Year festivities, and he rubbed shoulders with China’s elite—advising Premier Li Qiang on the 2024 work report and joining a Xi Jinping symposium in February.
Philosophy: Leading, Not Following
In a July 2024 interview, Liang laid bare his worldview: China must stop “freeriding” on Western tech and lead in originals. He critiques imitation (e.g., copying outdated models like Llama) and pushes for curiosity-driven innovation over capital alone. AGI, he predicts, is 2–10 years away, with DeepSeek betting on math/code, multimodality, and efficient reasoning to close China’s fourfold resource gap. Open-sourcing everything builds ecosystems and talent; closed models only delay rivals. No IPO or fundraising for him—DeepSeek stays research-focused, offering cheap APIs while staying profitable. As he puts it, “We’re done following. It’s time to lead.”
Liang’s story is one of quiet persistence: from rural math whiz to quant billionaire to AI trailblazer, proving startups can disrupt with smarts, not just scale.
Who Is Liang Wenfeng, the Founder of AI Disruptor DeepSeek?
Liang Wenfeng - Wikipedia
Meet DeepSeek founder Liang Wenfeng, a hedge fund manager
Interview with Deepseek Founder: We’re Done Following
What’s the story of DeepSeek and its founder Liang Wenfeng?