# Yao Fu

> Yao Fu (符尧) is an AI researcher at xAI specializing in large language model reasoning, efficient inference, and distributed systems. A PhD graduate of the University of Edinburgh, he previously worked at Google DeepMind on Gemini 3 and Project Astra. With over 5,000 citations and key papers like ServerlessLLM (OSDI '24) and DuoAttention (ICLR '25), Fu bridges systems engineering and ML research. He writes the 'Yao Fu' newsletter on Notion and is known for the Chain-of-Thought Hub benchmark repository, which helped track LLM reasoning progress across the field.

- **Role:** Researcher at xAI
- **Organizations:** xAI, University of Edinburgh, Google DeepMind, NVIDIA, Columbia University, Peking University
- **Nationality:** Chinese
- **Education:** PhD, Computer Science, University of Edinburgh, M.S., Columbia University, B.S., Computer Science and Technology, Peking University
- **Known for:** 5,000+ citations on Google Scholar across 18 papers, 209 highly influential citations on Semantic Scholar, ServerlessLLM accepted at USENIX OSDI 2024 - achieves 6-10x faster model loading

## Career timeline

- **2017** — Awarded CCF Elite Collegiate Award at Peking University
- **2018** — Awarded Wangxuan Scholarship at Peking University
- **2020** — Started PhD at University of Edinburgh, supervised by Prof. Luo Mai
- **2022** — Research internships at Allen Institute for AI and Alibaba Group
- **2023** — Presented 'Specializing Smaller Language Models towards Multi-Step Reasoning' as oral at ICML 2023; published Chain-of-Thought Hub benchmark
- **2023** — PhD completed at University of Edinburgh
- **2024** — ServerlessLLM paper published at USENIX OSDI '24
- **2024** — Joined Google DeepMind, worked on Gemini 3 perception and Project Astra
- **2024** — DuoAttention work submitted to ICLR 2025
- **2025** — Joined xAI as scaling researcher
- **2025** — Published comprehensive survey on LLM reasoning frontiers (arXiv 2504.09037)

## Achievements

- 5,000+ citations on Google Scholar across 18 papers
- 209 highly influential citations on Semantic Scholar
- ServerlessLLM accepted at USENIX OSDI 2024 - achieves 6-10x faster model loading
- DuoAttention accepted at ICLR 2025 - reduces LLM memory usage by up to 2.55x
- Chain-of-Thought Hub repository with 2,700+ GitHub stars
- ICML 2023 oral presentation for multi-step reasoning distillation paper
- CCF Elite Collegiate Award (2017)
- Wangxuan Scholarship, Peking University (2018)
- Contributed to Google DeepMind's Gemini 3 and Project Astra

## Latest updates

- **2025-04** — Published comprehensive survey 'A Survey of Frontiers in LLM Reasoning: Inference Scaling, Learning to Reason, and Agentic Systems' (arXiv 2504.09037)
- **2025-01** — Joined xAI as scaling researcher
- **2025-01** — Shared insights on Twitter about string-matching binary rewards and DeepSeek R1/K1.5 training approaches
- **2025-01** — DuoAttention paper accepted at ICLR 2025
- **2024-01** — ServerlessLLM paper accepted at USENIX OSDI 2024

## Links

- Website: https://franxyao.github.io/
- LinkedIn: https://www.linkedin.com/in/yao-fu-281847b5/
- Twitter/X: https://x.com/Francis_YAO_
- GitHub: https://github.com/FranxYao

---

Profile page: https://yespress.io/yao-fu
Published by YesPress — https://yespress.io
Last updated: 2026-04-17