Graduate researcher · Seoul, Korea
Mingyu Kim studies how models learn.
Statistics, language models, and causal reasoning for AI systems we can understand and trust.
I am a graduate student in Statistics and Data Science at Yonsei University. My work connects statistical inference with modern machine learning, with a current focus on diffusion language models and causal machine learning.
Open to research conversations and collaborations.
Questions I am
thinking about.
Across language models and causal systems, I am interested in representations, mechanisms, and reliable uncertainty.
Representation Learning in DLLMs
Causal Machine Learning
Pre-trained PFNs for Causal Effect Estimation
Mechanistic Interpretability
Conformal Inference in LLMs
Peer-reviewed
publications.
Recent work spanning causal mechanism shifts and post-acute sequelae of SARS-CoV-2.
ICML 2026 · Peer-reviewed
Dissecting Causal Mechanism Shifts via FANS: Function and Noise Separation
FANS separates changes in causal functions from changes in noise, offering a unified view of causal mechanism shifts across environments.
Proceedings of the 43rd International Conference on Machine Learning
IJID 2026 · Peer-reviewed
Static and Dynamic Scoring Systems for Post-acute Sequelae of SARS-CoV-2 in a Korean Cohort
This work develops static and dynamic scoring systems to quantify post-acute sequelae of SARS-CoV-2 in a Korean cohort.
International Journal of Infectious Diseases
Research into
practice.
Selected projects across public health, natural language processing, and educational AI.
Estimation and Quantification of Long COVID in a Korean Cohort
Statistical research funded by Korea's Ministry of Food and Drug Safety.
Mitigation of Translationese with Synthetic Data
A style-transfer system that rewrites translated Korean more naturally while preserving meaning.
English-specified Educational Chatbot under CEFR
An industry-funded educational chatbot designed around CEFR-aligned English learning.
A statistical
point of view.
Education and experience built around careful inference, collaborative research, and applied problem solving.
Education & experience
Graduate Studies, Statistics & Data Science
Yonsei University, Seoul, Korea.
Statistical Consultation Researcher
Institute of Data Science, Yonsei University. Consultations included New Testament studies and environmental finance.
Undergraduate Research Assistant
MLAI Lab with Prof. Kyungwoo Song; reviewed work on LLM reasoning, causal inference, and sharpness-aware minimization.
B.S. in Applied Statistics
Yonsei University · GPA 4.25/4.3 overall, 4.27/4.3 major · Graduated with highest honor.
Beyond the lab
Teaching
Student Tutor, Deep Learning (STA3140), Spring 2024 · Course Tutor, SW Programming (YCS1002), Fall 2021.
Awards
Guwon Scholarship · Yonsei University Academic Award, Top 1% of 120+ students.
Academic communities
14th member, Yonsei Artificial Intelligence · 12th member and Head of Academic Team, Yonsei Data Science Lab.
Tools & languages
R and Python · Deep-learning libraries · Java · Korean, English, and Japanese · ADsP certified.
Additional
Completed military service as a sergeant, August 2022–February 2024.