CV

A brief overview of my background. The full CV is available as a PDF.

Education

• M.S. in Statistics and Data Science, Yale University, Aug 2025 – Dec 2026 (expected)
• B.S. in Computer Science, Nanjing University, Sept 2021 – Jul 2025
  • GPA: 4.51/5.0 (Rank: 3/103)
• Exchange Student, Berkeley International Study Program, University of California, Berkeley, Jan 2024 – May 2024
  • GPA: 4.0/4.0

Professional Experience

• Research Intern (AI/ML), Snowflake AI Research, Bellevue, WA, May 2026 – Aug 2026
  • Manager: Zhewei Yao, Mentor: Boyi Liu
  • LLM post-training and agentic RL; developed SWARM, multi-agent RL training for deep research (continued jointly with Yale).
  • Explored RL algorithms for multi-agent training, including GRPO, DPPO, CISPO, and SAO.

Research Experience

• Research Assistant, Yale University, Aug 2025 – Present
  • Supervisor: Prof. Zhuoran Yang, Department of Statistics and Data Science
  • LLM pre-training and post-training, agentic RL, and evaluation of LLM agents.
  • Explored Nanotron for LLM pre-training and conducted LLM post-training with RL frameworks such as VERL and SLIME.
  • Contributed to freephdlabor, an open-source multiagent framework that automates the full research lifecycle.
  • Developing ForensicsBench and AD-Bench, benchmarks evaluating LLM agents on digital forensics and Attack-and-Defense CTF.
• Student Intern, Yale University, Jul 2024 – Jan 2025
  • Supervisor: Prof. Zhuoran Yang, Department of Statistics and Data Science
  • Mechanistic interpretability: identified the kernel estimator learned by one-layer Transformers for linear regression, and characterized the training dynamics and learned patterns.

See Projects for project details.

Publications

Build Your Personalized Research Group: A Multiagent Framework for Continual and Interactive Science Automation

Ed Li, Junyu Ren, Xintian Pan, Cat Yan, Chuanhao Li, Dirk Bergemann, Zhuoran Yang. "Build Your Personalized Research Group: A Multiagent Framework for Continual and Interactive Science Automation." arXiv preprint arXiv:2510.15624, 2025.

In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention

Jianliang He, Xintian Pan, Siyu Chen, Zhuoran Yang. "In-Context Linear Regression Demystified: Training Dynamics and Mechanistic Interpretability of Multi-Head Softmax Attention." arXiv preprint arXiv:2503.12734, 2025.

Awards & Honors

• First Class People’s Scholarship, Nanjing University (Top 5%)
• Second Prize Scholarship for Basic Subjects, Nanjing University

Skills

• Languages: Python, C/C++, Java, TypeScript, SQL, LaTeX
• Training & RL: PyTorch, Megatron, Nanotron, PRIME-RL, VERL, SLIME
• Inference & Agents: vLLM, SGLang, LiteLLM
• Infrastructure: Linux, Docker, Kubernetes, Git, AWS, Weights & Biases