How Liang Wenfeng Built a World Class AI Model in a Small Budget
[HPP] Liang WenfengMarch 25, 20258 min
20 connections·24 entities in this video→Early Life and Academic Foundations
- 💡 Liang Wenfeng was born in 1985 in a small village in rural China, showing an early spark of intellectual curiosity despite humble beginnings.
- 📚 He pursued his academic journey at Zhejiang University, majoring in electronic information engineering and later earning a master's degree in information and communication engineering.
- 🔬 His master's thesis focused on research on target tracking algorithm, demonstrating his early drive to understand technology's intricacies.
Pioneering AI in Finance
- 📈 Between 2008 and 2014, Liang fearlessly explored quantitative trading, leveraging AI to navigate financial markets, even during the 2008 global financial crisis.
- 🤝 He co-founded Jacob Investment in 2013 and Highflyer Quantitative Investment Management in 2015, both using AI-driven trading models to manage billions.
- 🛠️ Liang's commitment to innovation was unwavering; he was actively involved in building and improving the trading systems by writing code daily.
Founding DeepSeek for AGI
- 🚀 In May 2023, Liang made the momentous decision to leave Highflyer and establish DeepSeek, an AI firm with the audacious goal of achieving Artificial General Intelligence (AGI).
- 🎯 His focus was on fundamental AGI research and innovation, prioritizing groundbreaking development over immediate commercial success.
- 🌱 DeepSeek's unconventional approach included hiring fresh graduates and young talents, believing in their potential over extensive experience.
Disrupting the AI Landscape
- 🧠 DeepSeek developed the innovative hybrid multi-head latent attention (MLA) architecture for its models.
- ⚡ In December 2024, DeepSeek released DeepSeek V3, demonstrating that cutting-edge AI could be developed with a fraction of the resources used by Silicon Valley giants.
- 🌐 Just weeks later, in January 2025, DeepSeek unveiled DeepSeek R1, a powerful 671 billion open-source AI model, built with only 2,048 Nvidia A800 GPUs at a cost of $5.6 million.
- 🏆 DeepSeek's success was highlighted when it surpassed ChatGPT to become the number one free app on the US iOS App Store.
Vision and Impact
- 🌍 Liang is committed to open-sourcing all his models, aiming for China to be a contributor rather than a
Knowledge graph24 entities · 20 connections
How they connect
An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.
Hover · drag to explore
24 entities
Chapters4 moments
Key Moments
Transcript29 segments
Full Transcript
Topics15 themes
What’s Discussed
Liang WenfengDeepSeekArtificial Intelligence (AI)AI ModelsArtificial General Intelligence (AGI)Quantitative TradingHedge FundsNvidia A800 GPUsOpen Source AIChatGPTHybrid Multi-Head Latent Attention (MLA)InnovationStartup FundingTechnology DevelopmentSilicon Valley
Smart Objects24 · 20 links
People· 5
Companies· 7
Products· 5
Locations· 2
Concepts· 4
Media· 1