🧬 PAST, our multimodal single-cell foundation model for histopathology and spatial transcriptomics in cancer, has been accepted in principle by Cell Systems. Code is available.
Recent News
More news →🎉 TRACE-GRPO: Teacher-Refined Advantage Credit Estimation for Long-Horizon Agentic RL has been accepted at EMNLP 2026. Our work uses teacher feedback to improve credit assignment for long-horizon LLM agents. Code is available.
🏅 Honored to receive the ICML 2026 Silver Reviewer Award. Grateful for the recognition and for the opportunity to contribute to the research community.
About Me
I am a Principal Research Scientist at the Shanghai Academy of Artificial Intelligence for Science (SAIS). Before joining SAIS, I spent eight years at Ant Group and Alibaba Group working primarily on computer vision and multimodal AI. Earlier, I held research and engineering roles at Sony in Stuttgart and Tractable in London.
My current research centers on building AI models that connect biological understanding with the design of programmable therapeutics. I develop gene and RNA foundation models for biological representation learning and methods for sequence- and structure-based RNA design. Building on this foundation, I explore how AI scientists and autonomous labs can use these models to formulate hypotheses and guide experiments. New experimental data and insights can then feed back to improve the foundation models, closing the loop between modeling and experimentation. This effort includes contributions to NovaInspire (星河启智), SAIS's open platform for scientific discovery.
After more than a decade in AI, I see today's advances as a turning point for science. AI is accelerating how we learn and connect ideas across disciplines, opening new ways to tackle problems once considered beyond reach.
For example, AI may create unprecedented opportunities to bring us closer to truly programmable and personalized medicine. Together with my collaborators, I aspire to realize this vision through nucleic acid therapeutics that reach targets conventional therapies struggle to address.
20+
papers
10+
international patents
10+
years in AI
Research Highlights
Gene & RNA Foundation Models
Learning biological representations with gene and RNA foundation models, and developing sequence-based siRNA design and structure-based RNA design methods.
AI Scientists
Reasoning, tool use, and learning over long-horizon tasks to support scientific discovery and experimental decision-making.
Autonomous Labs
Connecting molecular design with experiment planning, execution, and measurement in feedback loops that guide the next round of discovery.
Programmable Therapeutics
A long-term vision of using disease mechanisms to guide target selection and RNA design toward programmable and personalized nucleic acid therapies.
Selected Publications
View all publications →* Equal contribution · ✉ Corresponding Author
PAST: A multimodal single-cell foundation model for histopathology and spatial transcriptomics in cancer
Cell Systems, 2026 (accepted in principle)
GOLD: Geometric Optimized Latent Diffusion for Structure-Aware RNA Inverse Folding
NeurIPS 2026 (Spotlight)
Experience
Academic & Professional Activities →-
Apr. 2024 - Present
Principal Research Scientist
SAIS -
Apr. 2016 - Apr. 2024
Senior R&D Engineer
Ant & Alibaba Group -
Sep. 2012 - Nov. 2015
MSc in Communication Systems
EPFL -
Sep. 2008 - Jun. 2012
BEng with Honors
Zhejiang University -
Contact
I welcome collaborations in gene and RNA foundation models, sequence- and structure-based RNA design, AI scientists, autonomous labs, and programmable nucleic acid therapeutics.