I am a fourth-year Ph.D. student in Statistics at the University of Chicago, fortunate to be advised by Prof. Victor Veitch. My research focuses on the science of AI, with a particular emphasis on Large Language Model (LLM) interpretability. I am passionate about uncovering the underlying mechanisms of LLMs and contributing to the development of more transparent and interpretable AI systems. My work explores the geometric properties of language models, offering insights into their internal structures and decision-making processes.
Prior to my doctoral studies, I graduated from Seoul National University, majoring in Statistics and Mathematics. At SNU, I had the privilege of working with Prof. Sungkyu Jung.
I am committed to bridging the gap between theoretical statistics and practical AI applications, and I am always open to collaborations and discussions on AI interpretability, AI safety, and the future of AI technologies.
News
- 2025.01: Our paper, The Geometry of Categorical and Hierarchical Concepts in Large Language Models, is accepted to ICLR 2025 as an oral presentation!
- 2024.09: I started Research Intern at Netflix!
- 2024.07: I was at ICML 2024 to give a poster presentation at the main conference on The Linear Representation Hypothesis and the Geometry of Large Language Models. Additionally, I gave an oral presentation at Workshop on MI and a poster presentation at Workshop on TF2M, both on The Geometry of Categorical and Hierarchical Concepts in Large Language Models. I am very honored to receive the Best Paper Award at the MI Workshop! (main poster) (workshop poster) (slide)
- 2024.06: Our new preprint, The Geometry of Categorical and Hierarchical Concepts in Large Language Models, is released on arXiv! This is joint work with Yo Joong Choe, Yibo Jiang, and Victor Veitch.
- 2024.05: Our paper, The Linear Representation Hypothesis and the Geometry of Large Language Models, is accepted to ICML 2024!
- 2023.12: I was at NeurIPS 2023 to give an oral presentation at the Workshop on CRL. I presented our recent work on The Linear Representation Hypothesis and the Geometry of Large Language Models. (poster) (slide)
- 2023.11: Our new preprint, The Linear Representation Hypothesis and the Geometry of Large Language Models, is released on arXiv! This is joint work with Yo Joong Choe and Victor Veitch.
- 2022.10: I am honored to be a Takintayo Akinbiyi Memorial Award for Academic Excellence in Statistics recipient.
- 2021.09: I started PhD at UChicago!
- 2021.03: Our paper, Clustering on the Torus by Conformal Prediction, is accepted to Annals of Applied Statistics! This is joint work with Sungkyu Jung and Byungwon Kim.
Contact
Email: parkkiho@uchicago.edu