About Me

I am a Ph.D. student in the Department of Statistics and Data Science at Yonsei University, advised by Prof. Kyungwoo Song. I work on efficient and reliable foundation models, with a focus on large language models (LLMs): how to make them train more stably, run more efficiently, and remain reliable as data and environments change.

Research

My research follows two complementary directions:

• Efficient architectures and stable LLM training. I design model components and training methods that improve efficiency while preserving capability. My first-author work includes BHyT (ICML 2026), a bounded activation that replaces pre-layer normalization, and RRD (ICML 2026 Workshop), routing-and-residual distillation for converting dense LLMs into sparse mixture-of-experts models.
• Causal learning and robust generalization. I study how models can use causal structure to remain reliable under distribution shift and hidden confounding. My first-author work includes CCL (NeurIPS 2025), causal-aware in-context learning for out-of-distribution generalization, and CBPE (ACM SAC 2026), multimodal blood-pressure estimation with latent-confounder modeling.

I am currently extending these interests to LLM agents that generalize across changing environments, studying how causal knowledge learned in one environment can support decisions in another. My industry collaborations also include building LLM agents for vehicle fault diagnosis, with a focus on knowledge extraction, memory construction, and retrieval.

See my publications for the full list of papers and my CV for more on my research experience.

Education

• Yonsei University, Korea – Ph.D. in Statistics and Data Science (2024–2028, expected)
• University of Seoul, Korea – M.S. in Artificial Intelligence (2022–2024)
• Pukyong National University, Korea – B.S. in Electrical and Semiconductor Engineering (2014–2021)