Agentic evaluation & coordination
Protocol harnesses, multi-agent reasoning, critique uptake, failure recovery, and quality-cost evaluation.
People & collaboration
Team
Research grows through shared questions, careful experiments, and conversations across disciplines. Alongside my own research, I mentor students and work with research collaborators.
Explore a collaborationMentoring
Students & mentees
4 current students
I currently mentor 4 students. I welcome the curiosity and different perspectives that students bring to research.
Meet the people behind the research.
Student profiles and individual projects will be shared here soon.Working together
Research collaborators
I am open to collaboration with researchers, students, and teams interested in reliable AI agents and AI for Science. Current coauthors and their contributions can be explored through the research projects and publications.
Agent memory & learning
Controlled memory evaluations, message-level credit assignment, and evaluation-aligned training signals.
AI for Science
Experiment-grounded preference learning, DPO/PPO post-training, and generative or multi-fidelity methods for materials and molecules.
Open research resources
Code & datasets
I share code on GitHub and datasets through the AgentsSci organization on Hugging Face. Research descriptions and project information are collected on this personal website.
Get in touch
Have a question we could explore together?
Send a short introduction, the research question you are interested in, and any relevant work or links. For student research inquiries, please include your background and availability.