Yiming Yang

Yiming Yang is a Professor at the Language Technologies Institute and Machine Learning Department within the School of Computer Science at Carnegie Mellon University, where he has held faculty positions since 2003. His research spans foundational and applied aspects of machine learning, artificial intelligence, and scientific computing.

• Professor, Carnegie Mellon University (2003–Present)
• Associate Professor, Carnegie Mellon University (1996–2003)

Yang's research focuses on LLM-based problem-solving agents, combinatorial optimization, and scalable oversight frameworks. His work explores diffusion models, Langevin dynamics, and Fourier neural operators for NP-hard problems, while advancing reinforcement learning techniques for self-play supervision and principle-driven fine-tuning of large language models.

Recent publications highlight his contributions to code synthesis, PDE solving, and multi-agent reinforcement learning. Key methodologies include demonstration-guided control, retrieval-augmented reasoning, and test-time scaling laws. His team has developed frameworks like FEEDER for efficient in-context learning and μTransfer-FNO for zero-shot hyperparameter transfer in PDE solvers.

Notable scientific achievements include:

• Best Student Paper Runner Up (2013)
• Best Theoretical Paper Award (1994)
• Best Theoretical Paper Award (1993)

Yang has mentored over 20 PhD students and postdocs, including Shengyu Feng, Zhiqing Sun, and Aman Madaan, across domains like graph learning, extreme multi-label classification, and language model alignment.