This may be the person Lei Jun values the most at present.
Xiaomi recently released a promotion list internally, and Luo Fuli, the 31-year-old head of the large model team, was promoted to level 22. This means that it only took her just 10 months from her official announcement in November 2025 to reach the top of Xiaomi’s rank ceiling.
In this way, it set a record for the fastest promotion in Xiaomi’s history.
This girl from Yibin, Sichuan, graduated from Beijing Normal University with a bachelor’s degree and a master’s degree from Peking University. Her well-known experience is working at DeepSeek. She was once dubbed a “genius girl” by the outside world, and was eventually recruited by Lei Jun with a lot of money. But she has expressed to the outside world more than once that she is not a genius and just wants to do things quietly.
Zooming in, Luo Fuli’s exceptional promotion is the true epitome of talent competition in China’s AI industry. This group of young people is leading the trend of China’s AI industry.
Lei Jun recruits post-95 girls
Fastest promotion speed
In 10 months, Luo Fuli’s Xiaomi career was in full swing.
Along with the news of this promotion, the MiMo team officially released and open sourced the Xiaomi MiMo-V2.6 series, which includes two native full-mode models, Pro and Flash. Thanks to the expansion of RL computing power, MiMo-V2.6-Pro scored 46 points in the Artificial Analysis Intelligence Index (AA comprehensive intelligence index), ranking first in the open source weight model; and at the same intelligence level, the price is only 1/20 to 1/60 of overseas models.
This is regarded by the MiMo team as a key step in exploring the RSI (recursive self-improvement) path. Luo Fuli said frankly, “The research innovation and engineering challenges behind this work exceed DeepSeek R1, which I was partially involved in.”
According to public information, excluding Xiaomi founder Lei Jun, Xiaomi Group has ten levels of ranks ranging from 13 to 22, with directors corresponding to levels 19 to 20 and vice presidents corresponding to levels 21 to 22. The comprehensive annual salary of Xiaomi level 19 employees can reach more than 2 million yuan, which means that the salary of level 22 employees is bound to be higher. In other words, Luo Fuli has joined the ranks of Xiaomi’s core executives after leading the team to complete the launch of new models.
To trace the story of Luo Fuli joining Xiaomi, we have to start with a hot search.
Early last year, the Securities Times reported that Lei Jun had hoped to poach Luo Fuli with an annual salary of tens of millions, and the latter unexpectedly became popular. Then in November last year, she posted in her circle of friends “Intelligence will eventually move from language to the physical world”, announcing that she had joined the Xiaomi MiMo team.
Soon, Luo Fuli made her public debut on Xiaomi. At Xiaomi’s “People, Cars, and Homes Ecosystem” partner conference in December last year, she officially appeared as the person in charge of MiMo’s large model.
In March this year, Xiaomi released three models at once, including the flagship base model Xiaomi MiMo-V2-Pro for the agent era, as well as the full-modal base model Xiaomi MiMo-V2-Omni and the speech synthesis model Xiaomi MiMo-V2-TTS.
At that time, Luo Fuli posted on social media, “I call this a silent ambush not because we had a plan, but because the transformation from Chat to Agent paradigm happened so fast that even we ourselves could hardly believe it. Somewhere in the middle, there is a process that is exciting, painful, and fascinating.”
In July this year, Xiaomi adjusted the organizational structure of Xiaoai and split its technical capabilities into three parts: basic model, cloud and device. Among them, the basic model capabilities are provided by the MiMo team headed by Luo Fuli, which also shows that her business jurisdiction has been further expanded. Until this promotion, Luo Fuli reached the top of Xiaomi’s rank ceiling in just 10 months.
Lei Jun’s desire for young AI talents is self-evident.
Liang Wenfeng can’t keep her
Luo Fuli’s growth history
Luo Fuli’s death is not the legendary story of becoming famous at a young age as imagined by the outside world.
Born in an ordinary family in Yibin in 1995, Luo Fuli attended Yibin No. 1 Middle School in high school. Her class teacher once commented that she was “not top-notch, but hardworking enough.” Later, she was admitted to Beijing Normal University to major in computer science. In her own words, she was “an absolute ‘low-starter’ in the computer field.” Even so, Luo Fuli started preparing to study computer science at Peking University when she was a sophomore, and after hard work, she finally got her wish.
Initially, her research interests focused on word sense disambiguation, a subfield in natural language processing (NLP). But with the rise of large model technology, she judged that the essence of traditional disambiguation tasks is being covered by the “next word prediction” paradigm of large models. As a result, she decisively shifted her research focus to text generation, which also paved the way for her future development of large models.
The year when Luo Fuli really rose to prominence was in 2019, when she published 8 papers at ACL, the top international conference in the field of natural language processing, of which 2 were the first author. The label of “genius girl” also followed.
But Luo Fuli has an almost instinctive rejection of this label. She once bluntly stated in an interview with the Peking University Alumni Association that such labels are stereotypes created to attract attention. “It glosses over the true hard work, tenacity and dedication in the scientific research process, implying that success is innate.”
Recalling her years as a graduate student at Peking University, Luo Fuli once lamented, “I benefited a lot from Peking University’s free scientific research environment.” At the same time, Peking University also gave her the earliest “enlightenment of collaboration.” In the laboratory of Peking University, she experienced the power of teamwork for the first time: some students were good at building GPU clusters, while others specialized in algorithms, and everyone showed their strengths.
After graduating with a master’s degree, Luo Fuli chose Alibaba as the first stop of her career. She joined Alibaba Damo Academy through “Alibaba Star” and served as a researcher in the Machine Intelligence Laboratory, engaged in research in the field of artificial intelligence. Here, she led the development of the multi-language pre-training model VECO and promoted the open source work of AliceMind.
Three years later, Luo Fuli joined DeepSeek’s parent company Huanfang Quantitative to engage in deep learning-related work. She later served as a deep learning researcher at DeepSeek and participated in the development of DeepSeek-V2 and other models. With the emergence of DeepSeek’s large models, Luo Fuli began to appear more in the spotlight. She once concluded, “The experience at DeepSeek was a major turning point.”
Looking back, Luo Fuli’s growth path did not involve any shortcuts or luck, but was a typical result of long-term technical compound interest, and she has always adhered to a de-mythified self-perception. As she said, she is not a genius, and her starting point is actually very ordinary. “It is the curiosity to explore essential issues, good mobility, and a little bit of tenacity that refuses to admit defeat that have led me to where I am today step by step.”
AI大戰
The golden age of top students
A war for AI talent is quietly emerging.
It reminds people of the previous scene. In December last year, Tencent appointed former OpenAI researcher Yao Shunyu as its chief AI scientist. The latter was only 27 years old at this time. Subsequently, the Hunyuan Multimodal Model Department and the Large Language Model Department merged to form a Basic Model Department, which was uniformly managed by this young man.
ByteDance is also firing on all cylinders. At the beginning of this year, DeepSeek core researcher Guo Daya resigned, triggering the pursuit of a number of major Internet companies, and finally he chose to join the Byte Seed department. At the same time, ByteDance launched the “Top Seed” school recruitment plan to recruit 30 top fresh Ph.D.s from around the world, emphasizing that it will provide top-level salaries in the industry with no upper limit.
Baidu also joined this competition. In July this year, Sun Tianxiang, who was born in 1997, officially joined Baidu as the head of the basic model research and development department and also joined the Baidu Model Committee. This PhD in computer science from Fudan University is the first core developer of the ChatGPT-like open source large model MOSS, and was the first to propose the Model-as-a-Service (MaaS) concept internationally.
Before young people even leave campus, big companies are already rushing to offer them olive branches.
Tencent officially launched the “Qingyun Project” intern recruitment in March this year, opening up six major technical fields such as large language models, multi-modality, and agents, including no cap on salary policy; Dark Side of the Moon announced the launch of Kimi’s “Travel Plan” for top talents, which will grant company options to interns who have not yet graduated; MiniMax has also launched Top Talent talent plan, CEO Yan Junjie even publicly stated: For truly top talents, there will be no cap on salary and options, no qualifications, only how big of a problem you can solve.
As early as the end of 2025, Business Insider disclosed a set of data for a 4-6 month AI internship and short-term research project. The monthly salary has reached the range of 7,000-18,000 US dollars, equivalent to about 49,000-126,000 yuan in RMB. According to a report by LatePost, the annual salary for Byte Top Seed students will be about 1.5 million yuan in 2024, rising to 3 million to 5 million yuan in 2025. Some core positions will be open to 6 million yuan in 2026, doubling year after year.
There is no doubt that the domestic competition for AI talents has been fully upgraded. From the early stage of grabbing people with high salaries and involution of benefits, it has advanced to an all-round talent warehousing model with strategic decentralization, resource tilting, and unconventional ranks. This also indicates that young technical talents have become the core competitiveness of AI companies and influence the direction of the industrial structure.
The reference provided by the paths of Luo Fuli, Yao Shunyu and others to young people is not a “genius narrative”, but a simpler logic: continue to accumulate engineering capabilities in the right direction, make cutting-edge technologies practical, and then amplify their own value on a large enough platform.
Just as Luo Fuli gave advice to young people at the Zhiyuan Conference: Always maintain a desire for exploration and curiosity. “Extremely use cutting-edge AI tools to cultivate your own judgment and scientific research aesthetics in the process of continuous trial and error. In an era of rapid technological change, unique cognition, judgment and aesthetics are the core and most irreplaceable competitiveness of young people.”
In the final analysis, the competition in AI is never just a competition between models and computing power, but a competition in talent density.
Related Reading
- An AI movie without actors won the Dragon Award first2026-10-06
- Just now, three Nobel Prize-winning scientists used photoproteins to “disassemble” the brain.2026-10-06
- This National Day, the first batch of robot employees have taken up their posts2026-10-06
- Indian media: The so-called Made in India by Apple is just the final tightening of the screws!2026-10-05
