DeepSeek Founder Liang Wenfeng's Path from Math Prodigy to AGI Disruptor

It's Liang Wenfeng Again

A few days ago, the annual college entrance exam (Gaokao) just wrapped up.

Coincidentally, over the past couple of days I've come across stories about a top figure in China's AI circle and his own Gaokao experience from years ago on Xiaohongshu several times.

That's right, this figure is none other than the founder of DeepSeek: Liang Wenfeng.

Everyone knows DeepSeek is very impressive now.

What people don't know is that over twenty years ago, when DeepSeek founder Liang Wenfeng took his Gaokao, he was already in a league of his own.

Over twenty years ago, Liang Wenfeng scored an outstanding 806 points, becoming the top Gaokao scorer in his region.

This was reported by the media at the time and is well-documented.

When talking about Liang Wenfeng, the first impression people have is that he is low-key.

Although DeepSeek has released many versions, large and small, there has never been a product launch event, nor any market promotion.

Liang Wenfeng himself rarely makes public appearances. If you try to search for a few photos of him online, the results are always the same handful, so people have always been curious about his journey.

Therefore, in this article, we will systematically sort out Liang Wenfeng's academic and entrepreneurial journey, to see how he got to where he is today, step by step, becoming a top-tier entrepreneur in the AI field.

Note: Looking at other people's experiences can often give us some inspiration.

Academic and Entrepreneurial Journey

Public records show that Liang Wenfeng was born in 1985 in Zhanjiang, Guangdong, into a family of teachers; both his parents were primary school teachers.

Showing a natural talent for learning from a young age, Liang Wenfeng excelled academically and was particularly good at mathematics.

Upon graduating from primary school, Liang Wenfeng gained direct admission to Wuchuan No.1 Middle School through subject competitions, where he completed his junior and senior high school education.

In junior high, Liang Wenfeng's overall academic performance was not particularly outstanding because he was quite lopsided, with exceptional strength in mathematics and physics.

By senior high, his academic performance had improved qualitatively. Especially after joining the school's Olympiad training class, his grades jumped significantly, consistently placing him in the top three of his grade in his final year.

In 2002, Liang Wenfeng entered Zhejiang University's Electronic Information Engineering undergraduate program with excellent Gaokao scores, beginning his university journey.

Just how excellent were his Gaokao scores? As mentioned earlier, he scored 806 points, becoming the top Gaokao scorer in his region.

Although his score exceeded the admission threshold for Tsinghua University, he ultimately chose Zhejiang University's Electronic Information Engineering program, believing that ZJU's corresponding major and direction better matched his interests.

After completing his undergraduate degree, Liang Wenfeng continued his studies at ZJU, and in 2007, he was admitted to the Information and Communication Engineering graduate program at Zhejiang University, starting his postgraduate journey.

During his postgraduate studies at ZJU, Liang Wenfeng studied under Professor Xiang Zhiyu, focusing on computer vision research, and graduated with a master's degree from ZJU in 2010.

During his undergraduate and master's years at ZJU, Liang Wenfeng developed a strong interest in financial markets. Coupled with his excellent background in mathematics and science, this laid the groundwork for his future entrepreneurial path.

At the same time, many like-minded friends Liang Wenfeng met at ZJU later became important partners on his entrepreneurial journey.

Despite his strong academic background, Liang Wenfeng did not choose a traditional career path. Instead, he developed a keen interest in financial markets during university and astutely perceived the immense potential of quantitative investing through mathematics and artificial intelligence.

In 2008, during the global financial crisis, Liang Wenfeng assembled a team and began exploring the possibility of fully automated quantitative trading.

In 2013, Liang Wenfeng and his ZJU classmates co-founded Hangzhou Jacobi Investment Management Co., Ltd., a company focused on investment management and consulting. The company's name is said to be derived from the German mathematician Carl Jacobi, reflecting a love and respect for mathematics.

Two years later, in 2015, Liang Wenfeng and his founding team co-founded High-Flyer Quant, relying on mathematics and artificial intelligence for quantitative investment.

At that time, the founding team was full of high spirits and diligence, aspiring to become a world-class quantitative hedge fund.

And that year, Liang Wenfeng was just 30 years old.

As is well known, High-Flyer Quant is highly renowned in the quantitative investment field, once managing assets reaching hundreds of billions, and was a leader in the quantitative market at the time.

High-Flyer Quant has also consistently focused on the development and application of quantitative investment strategies centered on mathematics and artificial intelligence.

2016 was a major turning point for High-Flyer, as it officially achieved the AI transformation of its quantitative strategies, and the first stock portfolio generated by a deep learning algorithm model went live for real trading, using GPUs for computation. The company grew rapidly in the following years, with its asset management scale expanding quickly.

By 2021, High-Flyer's asset management scale surpassed the 100-billion-yuan mark, placing it among the top domestic quantitative private equity firms. Liang Wenfeng also officially became the low-key "Quant King" in industry circles.

That year, Liang Wenfeng was exactly 36 years old.

After achieving great success in quantitative investment, Liang Wenfeng turned his attention to the field of Artificial General Intelligence (AGI), an area with immense development potential, once again demonstrating his forward-looking insight. This led to the later story of DeepSeek.

In 2023, Liang Wenfeng officially founded DeepSeek, formally entering the field of Artificial General Intelligence (AGI).

DeepSeek is an innovative technology company focused on exploring the frontiers of AGI technology research and its practical applications.

DeepSeek's goal is not merely to develop efficient large language models, but to explore the essence of AGI and drive the progress of the entire ecosystem.

As its corporate culture states:

Dedicated to the cause of exploring the essence of AGI, not doing mediocre things, with curiosity, using the longest-term perspective to answer the biggest questions.

Liang Wenfeng mentioned in an interview that because large language models are a necessary path to AGI and have preliminarily shown characteristics of AGI, DeepSeek chose to start from here.

After DeepSeek was founded, although its R&D team was not large, it achieved major breakthroughs in just over a year, successively launching multiple models.

Does everyone still vividly remember the unprecedented sensation when DeepSeek-R1 was released? That was also a highlight moment for domestic AI, shocking Silicon Valley for a time.

Because from that moment on, the monopoly of AI large models by a few international giants was thoroughly broken. DeepSeek, with its open-source, efficient, and lower-cost technological innovation, truly promoted the affordability and democratization of AI large models.

And this is the greatest significance of DeepSeek's emergence.

Liang Wenfeng's 50 Top-Level Thoughts

Although Liang Wenfeng rarely makes public appearances and gives even fewer interviews,

back when DeepSeek was just born, he did participate in an exclusive interview with DarkSurge Waves, which I watched in its entirety at the time.

I must say, many of the viewpoints, technical insights, and industry understandings Liang Wenfeng mentioned in it are very inspiring to us.

We have shared this content before, and perhaps many of you have seen it. Today, let's specially sort out these 50 core thoughts of Liang Wenfeng from that time and share them with everyone.

Feel free to take a look and see if it brings you any new inspiration.

About What DeepSeek Does

About Understanding AI and the Industry

About Understanding Technology and Innovation

About Understanding Hiring and Evaluation

About Understanding Open Source

Final Words

Today, Liang Wenfeng's name has long resounded throughout the global tech world. At the same time, an AI wave set off by DeepSeek has already spread on a large scale worldwide.

Over the past two years, the pace of development in the AI field is evident to all, with the speed of technological iteration far exceeding expectations.

Perhaps in the not-too-distant future, AI will truly become an indispensable part of our daily lives, as essential as water, electricity, and gas.

DeepSeek's success has set a great example for the industry and is also a huge encouragement for domestic AI.

Today, the development of the domestic AI circle is becoming more and more prosperous. We believe there will be more iterative innovations in technology and products in the future, and we can look forward to that.

Note: This article is included in the GitHub open-source repository "Road to Coding" https://github.com/rd2coding/Road2Coding, which contains self-study roadmaps + knowledge point overviews for 6 major programming directions (positions), interview points, my resume, several hardcore pdf notes, as well as programmer life and reflections. Welcome to star.