My name is Shang Zhou, and I am a Ph.D. student in Computer Science at the University of California, San Diego (UCSD), fortunate to be advised by Professor Jingbo Shang. My research is centered on enhancing the capabilities of Large Language Models (LLMs), with a focus on their evaluation, efficiency, and controllability.

Beyond my academic pursuits, I have a deep background in competitive programming. I am an ICPC World Finalist (22nd Place, 2023) and a Codeforces International Grandmaster. This experience provides me with a unique perspective on algorithmic problem-solving and informs my work on pushing the boundaries of AI in complex reasoning tasks.

Research Interests

I am passionate about building more capable and reliable AI systems. My primary interests include:

• LLM Evaluation and Benchmarking: Developing robust frameworks to accurately assess the capabilities and limitations of LLMs on complex tasks, such as competitive programming and advanced reasoning.
• Efficient LLM Inference: Designing novel algorithms and strategies to reduce the computational cost of LLM inference while maintaining or even improving accuracy.
• Controllable Text Generation: Creating methods to precisely and smoothly control the attributes (e.g., style, formality, sentiment) of generated text.
• AI for Complex Problem Solving: Exploring the intersection of AI and competitive programming to enhance the reasoning and observational skills of intelligent agents.

News

• June 2025: Our work on LiveCodeBench Pro was featured in MIT Technology Review.
• May 2025: Our paper, “Scaling LLM Inference with Optimized Sample Compute Allocation,” was accepted to NAACL 2025.
• July 2024: Honored to serve as the President & ICPC Team Coach for the UCSD Competitive Programming Club.
• April 2024: Our team won the Champion title in the ICPC NSA Challenge.

Seeking Collaboration

I am actively seeking motivated and passionate students (undergraduate or master’s) to collaborate on research projects. If you are excited about AI and eager to publish at top-tier conferences, I would love to hear from you.

How to Reach Out: Please send me an email at shz060@ucsd.edu with the subject line [Collaboration Interest].

Please include in your email:

• Your CV (optional but recommended).
• A brief description of your research interests or specific problems you find exciting.
• A summary of your programming skills and any relevant project experience.

What I Offer:

• Guidance on cutting-edge research ideas.
• Mentorship in code implementation and experimental design.
• Access to necessary computational resources (compute power).
• Dedicated support throughout the paper writing and publication process.

Our goal is to produce high-impact work for publication at premier venues like NeurIPS, ICML, ICLR, ACL, NAACL, and EMNLP. I am committed to helping my collaborators achieve their academic and professional goals, including securing first-author publications for Ph.D. applications or gaining invaluable research experience.