I build practical AI agents for chemistry, materials workflows, and scientific automation.
我关注 AI 如何帮助科研工作流变得更可追溯、可复现、也更安全。
| Project | What it does | Built with |
|---|---|---|
| ChemBoMAS | A multi-agent workflow for validating chemistry data, recommending experiments with Bayesian optimization, and analyzing results. | Python, Google ADK, BayBE, RDKit |
| MatFlow Compiler | Compiles material-science papers, patents, and technical data into evidence-linked protocols, SOPs, safety reports, and recovery plans. | Python, structured data, scientific agents |
| welcome-to-sjtu | A Chinese campus-assistant skill that organizes study, course, academic, and everyday services for SJTU students. | Agent skills, information routing |
- 🧪 AI for scientific discovery and autonomous experimentation
- 📊 Bayesian optimization for chemistry and materials
- 🤖 Reliable multi-agent workflows with explicit evidence and safety boundaries
- 🛠️ Small tools that turn scattered information into useful, repeatable workflows
Python · Google ADK · BayBE · RDKit · pandas · scikit-learn · GitHub Actions
I care about clear problem definitions, runnable examples, traceable outputs, and honest project limitations. If one of these projects overlaps with your work, feel free to open an issue in the relevant repository.