Seonghwan Seo

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 scientific discovery, with a particular focus on small molecule drugs. I have developed deep learning models in various areas of drug discovery including generative modeling, virtual screening, property prediction, and pharmacophore modeling.

Recently, I have focused on generative modeling with Generative Flow Networks (GFlowNets). By incorporating synthesis-oriented generative modeling, I aim to replace traditional in silico virtual screening and in vitro high-throughput screening.

research highlights

Generative modeling

• Generative flow networks (GFlowNets): TacoGFN (TMLR 2024), CGFlow (in review)
• Synthesis-oriented molecular design: BBAR (Advanced Science 2023), RxnFlow (ICLR 2025)

In silico evaluation

• Protein-based pharmacophore modelings: PharmacoNet (Chemical Science 2024)
• Drug-likeness scoring: Unsupervised drug-likeness (Chemical Science 2022)