Chuancheng Shi

Hi! 👋 I am a researcher at the University of Sydney, where I am fortunate to be advised by Prof. Zhiyong Wang. Currently, I am also a Research Intern at the NExT++ Center, collaborating closely with Prof. Tat-Seng Chua and Research Fellow Fei Shen.

Research Interests 🧠

My research focuses heavily on the interpretability of large foundation models. I am deeply fascinated by the internal mechanisms of modern AI, and my current work investigates:

• Knowledge Composition: How internal knowledge structures are constructed and organized within neural networks.
• Neuron-level Representations: Unpacking the “atomic” units of model behavior and their semantic roles.
• Information Transformation: Understanding how massive models encode, organize, and transform complex information.

Vision & Goals 🔍🤝

My overarching goal is to uncover the fundamental principles that make AI systems more predictable, robust, and closely aligned with human intent. By deciphering the structural composition of knowledge, I aim to transform the “black box” of foundation models into transparent systems.

Ultimately, I want to help build AI that is transparent, controllable, and trustworthy. Beyond improving safety, I am excited about enabling richer, more reliable creative and reasoning capabilities through deep interpretability.

Thank you for visiting my website! 😊