Reasoning Systems
LLM reasoning, tiny recursive models, planning, and distillation methods for more reliable inference.
Euijin Baek
Ph.D. Student in Computing Science, University of Alberta
About
I am a Ph.D. student in Computing Science at the University of Alberta, advised by Prof. Mi-Young Kim and Prof. Randy Goebel. My research interests span reasoning in language models, planning and distillation for small neural reasoning systems, online learning, reinforcement learning, and robust decision-making under uncertainty.
Before my Ph.D., I completed an M.Sc. in Statistics at the University of Toronto and worked as a data scientist and ML engineer in recommender systems, NLP, and digital marketing platforms.
News
- July 2026 Our Findings of ACL 2026 paper, "TRM-Planner: Offline Target Planning and Distillation for Tiny Recursive Models" , is available on ACL Anthology.
- July 2025 Our UAI 2025 paper on corruption-robust variance-aware generalized linear bandits appeared in PMLR.
- Sep. 2024 I started my Ph.D. in Computing Science at the University of Alberta.
Research Focus
Online Learning
Robust bandit algorithms for sequential decision-making under heavy-tailed rewards and adversarial corruption.
Applied ML Systems
Experience building recommender systems, NLP pipelines, and production ML workflows with PyTorch, TensorFlow, Airflow, Docker, and AWS.
Publications
Findings of ACL 2026
TRM-Planner: Offline Target Planning and Distillation for Tiny Recursive Models
A two-stage teacher-cache distillation recipe for Tiny Recursive Models that shifts compute to offline target planning while leaving student-time inference unchanged.
Corruption-Robust Variance-aware Algorithms for Generalized Linear Bandits under Heavy-tailed Rewards
Variance-aware robust generalized linear bandits under heavy-tailed rewards and reward corruption.