About

I am a PhD student in Computer Science at the University of Hong Kong, advised by Prof. Ping Luo, and a research intern in the Efficient AI team at NVIDIA Research with Enze Xie and Prof. Song Han. I received a dual degree from Tsinghua University: a B.S. in Mathematics and Physics and a B.Eng. in Measurement & Control Technology and Instruments.

My research spans three directions: agents that improve AI systems on their own, together with the systems that make large generative models fast; efficient training and post-training of generative models; and robot learning.

News
  • Our video-acceleration task was selected for Scale AI RSI-Bench.
  • SoL-Pi released: auto-research loops that discover token-efficient agent harnesses.
  • Sol Video Inference Engine released on arXiv.
  • Sol-RL released: asymmetric scaling of diffusion RL with low-cost rollout proxies.
  • Joined NVIDIA Research as a research intern.
Direction 1

Agents & ML Systems

Agent loops that research and optimize AI systems on their own, and the serving and kernel work that makes large video models fast.

Sol Video Inference Engine: Agent-Native Full-Stack Acceleration Framework for Efficient Video Generation
Yitong Li, Junsong Chen, Haopeng Li, Haozhe Liu, Jincheng Yu, Ligeng Zhu, Ping Luo, Song Han, Enze Xie
arXiv 2606.23743, 2026
Paper Code Selected for Scale AI RSI-Bench 9.4k GitHub stars
SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
Haozhe Liu*, Tian Ye*, Sensen Gao†, Qihang Cao†, Yitong Li†, Mingchen Zhuge, Duomin Wang, Ruihua Zhang, Ping Luo, Jiawang Bian, Lei Zhu, Ligeng Zhu, Enze Xie, Song Han
arXiv 2609.20519, 2026
Paper Code 3.5k GitHub stars
Direction 2

Generative Model Training

Pre-training and post-training of image and video generators, with a focus on making reinforcement learning cheaper at scale.

Sol-RL: Asymmetric Scaling of Diffusion Reinforcement Learning via Low-Cost Rollout Policies
Yitong Li*, Junsong Chen*, Shuchen Xue*, Pengcuo Zeren, Siyuan Fu, Dinghao Yang, Yangyang Tang, Junjie Bai, Ping Luo, Song Han, Enze Xie
arXiv 2604.06916, 2026
Paper Project Code Up to 4.64× faster convergence
HunyuanVideo 1.5 Technical Report
Bing Wu, Chang Zou, Changlin Li, et al. (including Yitong Li)
arXiv 2511.18870, 2025
Paper
Captain Safari: A World Engine with Pose-Aligned 3D Memory
Yu-Cheng Chou, Xingrui Wang, Yitong Li, Jiahao Wang, Hanting Liu, Cihang Xie, Alan Yuille, Junfei Xiao
CVPR 2026
Paper
Vision-Language-Vision Auto-Encoder: Scalable Knowledge Distillation from Diffusion Models
Tiezheng Zhang, Yitong Li, Yu-Cheng Chou, Jieneng Chen, Alan L. Yuille, Chen Wei, Junfei Xiao
NeurIPS 2025
Paper
Direction 3

Robotics

Diffusion policies grounded by language-model code and learned dynamics, and manipulation in cluttered scenes.

CodeDiffuser: Attention-Enhanced Diffusion Policy via VLM-Generated Code for Instruction Ambiguity
Guang Yin*, Yitong Li*, Yixuan Wang*, Dale McConachie, Paarth Shah, Kunimatsu Hashimoto, Huan Zhang, Katherine Liu, Yunzhu Li
RSS 2025
Paper
Neural Dynamics Augmented Diffusion Policy
Ruihai Wu*, Haozhe Chen*, Mingtong Zhang*, Haoran Lu, Yitong Li, Yunzhu Li
ICRA 2025
Project
Broadcasting Support Relations Recursively from Local Dynamics for Object Retrieval in Clutters
Yitong Li*, Ruihai Wu*, Haoran Lu, Chuanruo Ning, Yan Shen, Guanqi Zhan, Hao Dong
RSS 2024
Paper
* Equal contribution. † Core contributor.
Experience
  • NVIDIA Research, Efficient AI Team · Research InternDec 2025 – nowWith Enze Xie and Prof. Song Han
  • Tencent Hunyuan, Video Foundation Model Team · Research InternSep 2025 – Dec 2025With Qi Tian and Zhao Zhong
  • Columbia University · Research Intern, Robotic Coding AgentsAug 2024 – Jan 2025With Prof. Yunzhu Li and Prof. Fei-Fei Li
Education
  • The University of Hong Kong · Ph.D. in Computer Science2025 – 2029
  • Tsinghua University · B.S. in Mathematics and Physics2021 – 2025