Han Fang is a Research Lead in Core LLM at Meta’s SuperIntelligence Labs, with research interests in reasoning and agents. In 2023-2024, he was a Senior Research Manager and led Meta AI’s post-training for two years. He debut Meta AI in 2023 and launched Llama 2 & 3 models into Family of Apps, improving model’s general quality, growing it to 1 billion MAU. He boostrapped the modeling team from 0 to 50+ people, initiated and led the following work-streams: alignment integration, core capabilities, tool use and orchestration, and data flywheel. In 2025, he focuses on teaching models to think, reason, and use tools. ​

Han holds a PhD in Applied Mathematics and has published in top-tier venues with 7500+ citations. He is a recipient of the President’s Award to Distinguished Doctoral Students, the Woo-Jong Kim Dissertation Award, and the Excellence in Research Award.

Google Scholar / CV / Linkedin / Twitter

News

• Meta AI reached MAU in Dec 2024 and in 1 billion MAU in May 2025.
  Improved Meta AI’s multilinguality, enabled the roll-out to 12 languages and 40+ countries. Blog Post / News (2025)

• Launched the voice mode and photo editing in Meta AI at Connect 2024
  Launched an updated Llama 3 model for voice mode. Improved Planner to enable photo editing with mutimodal inputs. Blog Post (2024)

• Launched Llama 3 on Meta AI and subsequently Llama 3.1.
  Developed Meta AI’s online RL with Mixture of Judges, improving reasoning, instructions following, safety, etc. Blog Post (2024)

• Launched Meta AI with an improved Llama 2 model.
  Debut Meta AI and launched Llama 2 into Family of Apps. Created Orchestrator for tool use, enabling search and image generation. Developed data flywheel for Reinforcement Learning from User Feedback. My talk at Meta’s Connect Conference (2023)

• Developed Meta AI Few-Shot Learner (FSL) that can adapt to new types of harmful content.
  Developed FSL which can work in 100+ languages, learns from images & text, and detects new forms of violations. Blog Post (2021)

• Training AI to detect hate speech in the real world
  Built a RL based framework to E2E optimize hate speech classifiers. Blog Post (2020)

Recent Papers

• Boosting LLM Reasoning via Spontaneous Self-Correction
  Xutong Zhao, Tengyu Xu, Xuewei Wang, Zhengxing Chen, Di Jin, Liang Tan, Zishun Yu, Zhuokai Zhao, Yun He, Sinong Wang, Han Fang, Sarath Chandar, Chen Zhu · COLM (2025)

• Learning Auxiliary Tasks Improves Reference-Free Hallucination Detection in Open-Domain Long-Form Generation
  Chengwei Qin, Wenxuan Zhou, Karthik Abinav Sankararaman, Nanshu Wang, Tengyu Xu, Alexander Radovic, Eryk Helenowski, Arya Talebzadeh, Aditya Tayade, Sinong Wang, Shafiq Joty, Han Fang, Hao Ma · ACL (2025)

• Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization
  Zishun Yu, Tengyu Xu, Di Jin, Karthik Abinav Sankararaman, Yun He, Wenxuan Zhou, Zhouhao Zeng, Eryk Helenowski, Chen Zhu, Sinong Wang, Hao Ma, Han Fang · ICML 2025

• Step-KTO: Optimizing Mathematical Reasoning through Stepwise Binary Feedback
  Yen-Ting Lin, Di Jin, Tengyu Xu, Tianhao Wu, Sainbayar Sukhbaatar, Chen Zhu, Yun He, Yun-Nung Chen, Jason Weston, Yuandong Tian, Arash Rahnama, Sinong Wang, Hao Ma, Han Fang · arXiv (2025)

• Improving Model Factuality with Fine-grained Critique-based Evaluator
  Yiqing Xie, Wenxuan Zhou, Pradyot Prakash, Di Jin, Yuning Mao, Quintin Fettes, Arya Talebzadeh, Sinong Wang, Han Fang, Carolyn Rose, Daniel Fried, Hejia Zhang · ACL 2025

• Multi-IF: Benchmarking LLMs on Multi-Turn and Multilingual Instructions Following
  Yun He, Di Jin, Chaoqi Wang, Chloe Bi, Karishma Mandyam, Hejia Zhang, Chen Zhu, Ning Li, Tengyu Xu, Hongjiang Lv, Shruti Bhosale, Chenguang Zhu, Karthik Abinav Sankararaman, Eryk Helenowski, Melanie Kambadur, Aditya Tayade, Hao Ma, Han Fang, Sinong Wang · arXiv 2024

• The Perfect Blend: Redefining RLHF with mixture of judges
  Tengyu Xu, Eryk Helenowski, Karthik Abinav Sankararaman, Di Jin, Kaiyan Peng, Eric Han, Shaoliang Nie, Chen Zhu, Hejia Zhang, Wenxuan Zhou, Zhouhao Zeng, Yun He, Karishma Mandyam, Arya Talabzadeh, Madian Khabsa, Gabriel Cohen, Yuandong Tian, Hao Ma, Sinong Wang, Han Fang · arXiv 2024