profile.png

Taewon Kim

I’m Taewon Kim, a Ph.D. candidate at KAIST AI advised by Prof. Sungsoo Ahn. My research sits at the intersection of AI and the natural sciences, where I aim to build foundation models that turn complex biological data into usable scientific insight. Technically, my recent focuses are 3D protein representation learning and genomic large language models.

LinkedIn | Google Scholar | GitHub | X

News

Apr 18, 2026 I’ll be going to Rio for ICLR 2026 :brazil:
Apr 18, 2026 DNAChunker has been accepted at MLGenX@ICLR 2026 as Oral Presentation
Jan 15, 2026 MELD has been accepted at ICLR 26’
Dec 15, 2025 SICL has been accepted at WACV 26’
Oct 15, 2025 New Preprint - DNAChunker is out

Publications

  1. Learning Flexible Forward Trajectories for Masked Molecular Diffusion
    Hyunjin Seo*, Taewon Kim*, Sihyun Yu, and 1 more author   |  *Equal contribution
    ICLR 26’ | Paper | Code | Website
  2. DNACHUNKER: Learnable Tokenization for DNA Language Models
    Taewon Kim, Jihwan Shin, Hyomin Kim, and 5 more authors
    ICLR 26’ Workshop Oral | Paper
  3. REBIND: Enhancing ground-state molecular conformation via force-based graph rewiring
    Taewon Kim*, Hyunjin Seo*, Sungsoo Ahn, and 1 more author   |  *Equal contribution
    ICLR 25’ | Paper | Code
  4. AdapTable: Test-Time Adaptation for Tabular Data via Shift-Aware Uncertainty Calibrator and Label Distribution Handler
    Changhun Kim*, Taewon Kim*, Seungyeon Woo, and 2 more authors   |  *Equal contribution
    NeurIPSW-TRL 24’ | Paper | Code
  5. NOTE: Robust Continual Test-time Adaptation Against Temporal Correlation
    Taesik Gong, Jongheon Jeong, Taewon Kim, and 3 more authors
    NeurIPS 22’ | Paper | Code