Personal Website | Google Scholar | Rednote

I am a final-year master's student at the School of Artificial Intelligence and Data Science, University of Science and Technology of China (USTC), advised by Prof. Tong Xu and Prof. Enhong Chen.

I am currently a research intern at 通义千问, where I focus on the infrastructure and algorithmic framework for agents and agent reinforcement learning. My work centers on building reliable, scalable systems that support the training, evaluation, and deployment of next-generation agentic models.

Previously, I worked on large language model reasoning and training systems at Baidu ERNIE (Star) and StepFun, gaining hands-on experience across pre-training, long-context modeling, post-training, and reasoning-oriented system design, along with large-scale distributed training at the scale of hundreds to over a thousand GPUs.

My research interests lie in agents, reinforcement learning, reasoning, and the infrastructure that connects these areas into practical and scalable intelligent systems. If you are interested in collaboration or discussion on these topics, feel free to contact me.

• Step2-mini & Step3: Cost-Effective Multimodal Intelligence [Website] [Technical Report] [arXiv]
  Contributed to the end-to-end development of state-of-the-art large language and multimodal reasoning models, scaling from tens to hundreds of billions of parameters with high efficiency.

• Nano Agent Team [GitHub]
  Built an experimental multi-agent collaboration framework based on a file-system blackboard model, enabling autonomous planning, role coordination, and human-in-the-loop execution for complex tasks.

• Awesome-Parallel-Reasoning [GitHub Repository]
  Maintaining a curated repository that tracks the rapidly evolving landscape of parallel reasoning and multi-agent collaboration research, tools, and best practices in the LLM era.

• M.S. in Artificial Intelligence and Data Science, University of Science and Technology of China (2023 - Present)
• B.S. in Artificial Intelligence and Data Science, University of Science and Technology of China (2019 - 2023)

_{The right card is auto-updated daily and includes public repositories I contributed to, including external repositories.}

Built with simplicity in mind.