About

I'm a research associate at Harvard Medical School, working in the Zitnik Lab with Prof. Marinka Zitnik in the Department of Biomedical Informatics. I also collaborate closely with Prof. Nils Gehlenborg and colleagues in the HIDIVE Lab. Previously, I completed my M.S. in Computer Science at Columbia University as part of the Columbia NLP Group, and my B.S. (Honours) at The University of Sydney with Prof. Zhiyong Wang.

My research interest focuses on how AI can move from producing plausible biomedical knowledge to making claims that survive biological scrutiny. I enjoy developing methods that integrate perturbations, omics, and literature evidence to reason about cellular state and disease mechanisms in immunology, pharmacology, and cell biology.

I am particularly interested in the backside of AI-assisted discovery: grounding predictions in mechanism, tracking conflicting evidence, estimating when a model should abstain, and prioritizing experiments that can change what we believe.

If anything sparks your interest, I would love to hear from you. Please don’t hesitate to reach out!

News

2026
Qworld accepted to COLM 2026.
89% expert-criteria coverage and 79% novel criteria validated by human experts.
ToolUniverse now integrates natively with Claude Science.
One prompt installs 2,500+ scientific tools and bundled research workflows directly inside the workbench, with no MCP server or plugin required.
Medea featured in Nature Methods.
With the Ideker Lab, we validated Medea on a previously unpublished 5,806 × 41 yeast E-MAP (238,046 gene pairs). It outperformed GPT-5 and GPT-5.2 by up to 24.3% on experimentally measured synthetic lethal interactions.
2025
Our CURE-Bench workshop at NeurIPS received submissions from 320+ teams working on precision therapeutics.

Research

Qworld evaluation criteria animation
Shanghua Gao*, Yuchang Su*, Pengwei Sui, Curtis Ginder, Marinka Zitnik
COLM 2026 (Conference on Language Modeling)
Medea paper Figure
Pengwei Sui*, Michelle Li*, Shanghua Gao, Wanxiang Shen, Valentina Giunchiglia, Andrew Shen, Yepeng Huang, Zhenglun Kong, Marinka Zitnik
bioRxiv preprint | ...
Shanghua Gao, Richard Zhu*, Pengwei Sui*, Zhenglun Kong*, Sufian Aldogom, Yepeng Huang, Ayush Noori, Reza Shamji, Krishna Parvataneni, Theodoros Tsiligkaridis, Marinka Zitnik
arXiv preprint | ...
YAC paper Figure
Devin Lange, Shanghua Gao, Pengwei Sui, Austen Money, Priya Misner, Marinka Zitnik, Nils Gehlenborg
In review for ACM CHI 2026 | VIS x GenAI Workshop @ IEEE VIS 2025
DQVis paper Figure
Devin Lange, Pengwei Sui, Shanghua Gao, Marinka Zitnik, Nils Gehlenborg
NeurIPS 2025: Conference on Neural Information Processing Systems
Multi-LLM Integration Figure
Zhenglun Kong*, Zheng Zhan*, Shiyue Hou, Yifan Gong, Xin Meng, Pengwei Sui, Peiyan Dong, Xuan Shen, Zifeng Wang, Pu Zhao, Hao Tang, Stratis Ioannidis, Yanzhi Wang
arXiv preprint
KokoMind Figure
Weiyan Shi*, Liang Qiu*, Dehong Xu, Pengwei Sui, Pan Lu, Zhou Yu
LICA Algorithm Animation
Meng Zhou*, Ziyu Liu*, Pengwei Sui, Yixuan Li, Yuk Ying Chung
NeurIPS 2020 | RL Theory Workshop @ ICML 2020

Experience

Harvard Medical School, Boston MA, United States

Research Associate, DBMI, 2024-Present
Harvard Medical School Logo

Insilico Medicine, Shanghai, China

Consultant Intern, Business Development, Spring 2024
Insilico Medicine Logo

Yale Engineering School, (remote from New York)

Graduate Student Researcher, Fall 2023
Yale Engineering Logo

Columbia Engineering School, New York, NY, United States

Graduate Student Researcher, Columbia NLP Group, Summer 2023
Columbia Engineering Logo

Education

Columbia University, New York, NY, United States

Master of Science in Computer Science, 2023
Columbia University Logo

University of Sydney, Sydney, NSW, Australia

Bachelor of Science (Honours) in Computer Science, 2021
University of Sydney Logo

Teaching

Back in my undergrad days at USyd, I was a teaching assistant for the following classes: