About Me

I am an undergraduate student at Peking University, currently pursuing a double major in Statistics (School of Mathematical Sciences) and Economics (National School of Development).

My research interests are grounded in Statistical Foundations of Modern Machine Learning and Operations Research. I am particularly fascinated by the synergy between these theoretical pillars and their applications in Deep Learning (DL), Reinforcement Learning (RL), and Large Language Models (LLMs).

I am actively seeking PhD opportunities starting in Fall 2028, with a focus on Statistics, Computer Science, or Operations Research.

Education

• Peking University, B.S. in Mathematics (SMS) & B.A. in Economics (NSD), Aug. 2023 – Jun. 2028 (Expected)
• University of Copenhagen, Exchange Student in Department of SCIENCE, Sep. 2025 – Jan. 2026
• University of Wisconsin–Madison, Student Research Assistant in Statistics & Computer Science, Jul. 2026 – Sep. 2026

Research Highlights

Task Vector Geometry and Context Compression

Advisor: Prof. Yiqiao Zhong (UW–Madison) | Apr. 2026 – Present

This project studies how sequence models compress task-relevant information into hidden states.

See the ongoing research on GitHub.

Active Pareto Set Identification from Binary Pairwise Feedback

Advisor: Prof. Zhimei Ren (UPenn) | May. 2026 – Present

This project focuses on developing active evaluation algorithms for identifying Pareto-optimal LLMs from noisy binary pairwise comparisons.

Algorithm Implementation for Partially Observed Dynamic Tensor Regression

Advisor: Prof. Yi Chen (HKUST) & Prof. Biao Cai (CityU) | Jul. 2025 – Feb. 2026

This project focused on developing efficient optimization frameworks for ultra-high-dimensional dynamic tensor regression tasks.

• POSTER Framework: Investigated and implemented the POSTER framework to handle complex random and block missing data patterns in dynamic tensor structures.
• Theoretical Analysis: Initiated convergence analysis of the non-convex alternating minimization framework. I am currently utilizing tools from High-Dimensional Probability to establish statistical guarantees and error bounds under structural sparsity constraints.
• Simulations: Designed robust simulation pipelines in R to evaluate tensor structure recovery and validate theoretical rates of convergence.

Selected Coursework & Notes

I take pride in crafting polished LaTeX lecture notes for courses that spark my curiosity, as well as developing unofficial solution sets for textbooks that lack publicly available answers.

• Statistical Thinking (Peking University Spring 2025)
  Instructor: [Prof. Wei Lin], Peking University [Lecture Notes (PDF)]

• Ongoing work: Applied Stochastic Processes (Peking University Press, in Chinese)
  Author: [Prof. Dayue Chen] and [Prof. Fuxi Zhang], Peking University [Solution Manual (PDF)]

Personal Interests

Beyond my academic pursuits, I am a passionate explorer of life’s diverse flavors and experiences:

• Culinary Arts & Mixology: I enjoy cooking and am an aspiring mixologist with a keen interest in the craft of cocktails and coffee.
• Photography & Travel: I love capturing the world through my lens; my recent journeys have taken me through the landscapes of Spain, Norway, Iceland, and Japan.
• Creativity & Leisure: In my spare time, I find joy in assembling Gumdam models and LEGO, exploring immersive video game and film worlds.

Follow me on xiaohongshu (Rednote) for more of my personal updates! Twist.

Contact

• Email: twistshan1218@gmail.com
• GitHub: Twist-Shan
• WeChat: Twist_SsLl