Junu Kim

News

Sep 2026

Started as a postdoctoral researcher at KAIST.

Aug 2026

Received my Ph.D. in Artificial Intelligence from KAIST.

Jul 2026

C-Reason is accepted to MLHC 2026.

Apr 2026

LayerNorm Induces Recency Bias in Transformer Decoders is accepted to ACL 2026 Findings.

Apr 2026

Started a research internship at the Institute of Foundation Models, Post-Training Team (K2-Horizon).

Selected papers all publications →

MLHC 2026

Enhancing LLMs' Clinical Reasoning with Real-World Data from a Nationwide Sepsis RegistryWe propose enhancing the clinical reasoning capabilities of LLMs by leveraging real-world clinical data.arXivCode

ACL 2026 Findings

LayerNorm Induces Recency Bias in Transformer DecodersWe discover that LayerNorm induces recency bias in Transformer decoders.arXivCode

MLHC 2024

General-Purpose Retrieval-Enhanced Medical Prediction Model Using Near-Infinite HistoryWe propose REMed, the first retrieval-based model for medical prediction tasks, which reduces the need for domain experts in the development process.PaperCode

ACL 2024 Findingsco-first author

Publicly Shareable Clinical Large Language Model Built on Synthetic Clinical NotesWe introduce the first publicly sharable clinical large language model, Asclepius.PaperCodeHugging Face

Experience & Education

Experience

Sep 2026 – present

Postdoctoral Researcher, KAISTAdvisor: Edward Choi

Apr – Aug 2026

Research Intern, Institute of Foundation ModelsPost-Training Team · Project: K2-Horizon · Manager: Haonan Li, Rupesh Srivastava

Jun – Dec 2025

Research Intern, Microsoft Research AsiaArtificial Intelligence & Reasoning Team · Project: LayerNorm Induces Recency Bias in Transformer Decoders · Mentor: Xiao Liu

Education

2022 – Aug 2026

Ph.D. in Artificial Intelligence, Kim Jaechul Graduate School of AI, KAISTAdvisor: Edward Choi

2018 – 2022

B.S. in Computer Science, School of Computing, KAIST