Seonghwan Seo

Korea Advanced Institute of Science and Technology (KAIST)

prof_pic.jpg

I am a Ph.D. student in the Department of Chemistry, KAIST, under the supervision of Prof. Woo Youn Kim.

My research area is AI-driven drug discovery, focusing on deep generative models for molecular design and in silico evaluation. During my Ph.D., I have specialized in integrating RL agents with autonomous laboratories to accelerate the discovery cycle.

More recently, I served as the engineering lead of Team KAIST for biomolecular structure prediction, developing AtlasFold for proteins and K-Fold for general biomolecular complexes. I am also extending my research into computational protein design, collaborating with Prof. Wengong Jin.

research highlights

Biomolecular structure prediction

Generative modeling

In silico evaluation

news

Oct 26, 2026 Our K-Fold preprint, “Predicting conformational changes induced by biomolecular interactions,” is now available on bioRxiv. Code: K-Fold.
Oct 01, 2026 Our preprint on synthesis-aware generative design in trillion-scale chemical spaces for automated drug discovery is now available on bioRxiv. Code: RxnFlow.
Jul 10, 2026 I organized a workshop on “The 2026 workshop on Generative AI and Agentic AI for Biology” (GenBio) at ICML 2026, Seoul, Korea.
Mar 02, 2026 2 papers are accepted on ICLR 2026 GEM Workshop: Boltz2ESi and AF3-MSD/AF3-Cluster.
May 02, 2025 1 paper is accepted to ICML 2025: CGFlow

selected publications

  1. bioRxiv
    Predicting conformational changes induced by biomolecular interactions
    Yeongnam Bae, Jinheon Baek, Eugene Choi, Jeong Hun Choi, Hyeongyu Han, Hyosoon Jang, Woosung Jeon, Jaehyeong Jo, Dongki Kim, Hyeongwoo Kim, Jun Hyeong Kim, Jun Hyeong Kim, Shinwoo Kim, Taewon Kim, Youngchan Kim, Joongwon Lee, Junyup Lee, Seokhyun Moon, Minha Park, Calvin Samuel, Hyunjin Seo, Jisu Seo, Seonghwan Seo, Mingyeong Shin, Yunsu Shin, Wonho Zhung, Sungsoo Ahn, Dongsu Han, Sung Ju Hwang, Ho Min Kim, Woo Youn Kim, Gyu Rie Lee, and Byung-Ha Oh
    bioRxiv, 2026
    Authors listed in alphabetical order; served as Engineering Lead.
    DOI HTML Code
  2. ICML
    Compositional Flows for 3D Molecule and Synthesis Pathway Co-design
    Tony Shen*, Seonghwan Seo*, Ross Irwin, Kieran Didi, Simon Olsson, Woo Youn Kim, and Martin Ester
    In International Conference on Machine Learning, 2025
    * Previously presented at ICLR AI4Mat and GEM Workshop 2025
    Award arXiv
  3. ICLR
    Generative Flows on Synthetic Pathway for Drug Design
    Seonghwan Seo, Minsu Kim, Tony Shen, Martin Ester, Jinkyoo Park, Sungsoo Ahn, and Woo Youn Kim
    In International Conference on Learning Representations, 2025
    * Previously presented at NeurIPS AIDrugX Workshop 2024
    arXiv Blog Code
  4. TMLR
    TacoGFN: Target-conditioned GFlowNet for Structure-based Drug Design
    Tony Shen, Seonghwan Seo, Grayson Lee, Mohit Pandey, Jason R Smith, Artem Cherkasov, Woo Youn Kim, and Martin Ester
    Transactions on Machine Learning Research, 2024
    arXiv Code
  5. Adv. Sci.
    Molecular generative model via retrosynthetically prepared chemical building block assembly
    Seonghwan Seo, Jaechang Lim, and Woo Youn Kim
    Advanced Science, 2023
    IF: 15.1
    Selected arXiv HTML Code