Bumjin Park (call me Leo)

I study the mind of AI through rigorous analysis of neural representations — the computational brain of artificial intelligence.

My research focuses on Interpretability, Cognitive Architectures, and Neural Representation — exploring how internal structures of AI models reveal the underlying principles of cognition shared between humans and machines.

Ultimately, I aim to develop neural reasoning systems that integrate human and computational forms of reasoning, advancing toward a unified form of general intelligence.

News

• October 1, 2025 — Started my three-month stay as a Visiting Scholar at NYU!
  If you’re interested in discussing research ideas, feel free to reach out.

• September 15, 2025 — My paper on LLM Jailbreak has been accepted at CKIM 2025 HCAI Workshop! Excited to share our findings on trustworthy and safe AI.

Education

Ph.D. Student in Artificial Intelligence (09/2023 – Present)

• Korea Advanced Institute of Science and Technology (KAIST), AI Graduate School
• Topic: Integrating Cognitive Architectures into Large Language Models 📂 Drive

M.S. in Artificial Intelligence (08/2023)

• Korea Advanced Institute of Science and Technology (KAIST), AI Graduate School
• Thesis: Partitioned Channel Gradient for Reliable Saliency Map in Image Classification 📂 Drive

B.S. in Mathematics (08/2020)

• Chung-Ang University, Korea
• Double Major in Software Engineering

🗂️ CV(October 16, 2025) | 📧 Email: bumjin@kaist.ac.kr | bumjini42@gmail.com