Luo Fuli is a 29-year-old researcher whom netizens and co-workers in China gave the nickname “AI prodigy.” She is known for her pivotal role in the development of DeepSeek-V2. This is China’s first artificial intelligence (AI) language model that could go toe-to-toe with OpenAI’s ChatGPT. According to the South China Morning Post, the large language model (LLM) was launched by DeepSeek on December 26, 2024, and it was trained with much fewer resources than Meta’s Llama. In a May 2023 interview with Chinese media outlet 36Kr, DeepSeek founder Liang Wenfeng said that when recruiting talent, the company prioritizes ability over experience. Local news sources suggest the team of developers at the startup consists of mostly graduates and university students. “Our core technical roles are filled with mostly fresh graduates or those with one or two years of working experience,” he explained. This strategy has helped DeepSeek to build a team of ambitious young researchers, including Gao Huazuo and Zeng Wangding. The two have been credited with key innovations in MLA architecture. Luo Fuli: Taking a chance at computer science paid off Among the most well-sought devs in the company is Luo Fuli. She is said to be a “brainy” and heartfelt tech expert with a strong background in natural language processing (NLP). Fuli was reportedly brought up in a modest environment. It is said that her interest in tech might have stemmed from her father, an electrical engineer. Luo’s rise in the AI field began at Peking University’s Institute of Computational Linguistics. Unconfirmed reports from deep dives and social media state she was initially uncertain about studying computer science. Fuli even failed a few times while at it. However, she eventually found her passion for AI and made a name for herself through groundbreaking research. Moreover, it is rumored that Luo received job invitations and offers while she was still in school, but she turned them all down. In 2019, she caught the attention of the Chinese tech sector after publishing eight papers about NLP at the Association for Computational Linguistics (ACL) conference. Her knowledge and input in NLP saw her receive several offers from major technology firms, particularly Alibaba. At Alibaba’s DAMO Academy, Luo contributed to VECO, a multilingual AI model. She worked on the company’s open-source AliceMind project, helping the online marketplace advance its AI initiatives. However, her ambitions grew beyond corporate research. To that end, she joined a role at Wengfeng-led DeepSeek AI in 2022 as a principal researcher. Luo Fuli and the young team of developers at DeepSeek AI At the very start, Luo was part of the team that actualized DeepSeek-V2. This is a cost-effective large language model that locals nicknamed “AI Pinduoduo”—a reference to the Chinese online e-commerce giant known for its affordable pricing. Feminism with Chinese characteristics. This is Luo Fuli, a prodigy at DeepSeek and author of 8 AI papers! She got her Masters degree from Peking University in 2020. Worked at Alibaba, joined DeepSeek in 2022, did amazing stuff, and now has been “stolen” by Xiaomi AI lab! pic.twitter.com/MCz3ahXKVJ — S.L. Kanthan (@Kanthan2030) January 29, 2025 Speaking at a tech conference in 2023, Luo brought to light how the model offers top-tier Chinese language capabilities. This rivaled even the best global AI systems the likes of ChatGPT and Qwei. Luo Fuli attributed DeepSeek-V2’s success to a combination of innovative architecture, robust infrastructure, and the company’s commitment to transparency. During her time at the company, DeepSeek openly shared its technical reports, model weights, and inference code on GitHub. This was aimed at actualizing its open-source approach to AI development. DeepSeek AI – The open-source technology at its peak One of DeepSeek-V2’s standout features is its use of Multi-Level Attention (MLA) and Mixture of Experts (MoE) architecture. MLA enables the model to focus on different levels of textual information—akin to how a human reader processes a book. It shifted attention between sentences, paragraphs, and chapters depending on context. Meanwhile, MoE optimizes computational efficiency by directing tasks to selected virtual “experts” within the model. The build-up reduced resource consumption while growing performance. Luo Fuli believes that China needs more AI labs. She insists her country should focus on practical and large-scale engineering projects. Luo Fuli has also been a strong advocate for a shift toward research that prioritizes real-world applications. The new direction should ensure that AI advancements translate into tangible benefits for businesses and consumers. The 29-year-old techie’s growing influence in the AI industry has not gone unnoticed. Per reports from the SCMP, Xiaomi’s founder personally offered her an annual compensation package of 10 million yuan, but it is unclear if she accepted the offer. Cryptopolitan Academy: FREE Web3 Resume Cheat Sheet - Download Now
