My path never followed a pre-planned route. From renewable-energy systems modelling, to materials and machine learning, to today's AI for Science, research platforms, and AI products — what I increasingly know about myself is that I love understanding complex problems, building models, and then making the system with my own hands.
I will enter a new field and try to genuinely understand its basic problems.
Rather than being defined by a single disciplinary tag, I care about the questions I keep working on.
Research, engineering, and products are one continuous growth path to me.
I want to keep the ability to learn fast and build things with my own hands.
A path from energy systems modelling to AI for Science, stressing the continuity of "problem – method – system", with school names not as the narrative axis.
Started research on electrolyser-array optimisation for hydrogen production: an electro-thermal coupled model, scheduling optimisation, and a control interface in Python.
Representative results: led a national-level undergraduate innovation project; Cup-AWE V1.0 software copyright.
Competed in national-level competitions around low-carbon technology, CCUS, and energy saving, while steadily strengthening computational modelling, data analysis, and research communication.
Representative results: three National Scholarships for undergraduates (three consecutive years, ranked first in major); multiple national-level competition awards.
Research focus shifted to molecular machine learning and quantum chemistry: a tiered PM6 → DFT → TDDFT/Multiwfn computation pipeline and an AI screening workflow.
Representative results: ~26K molecular-scale data production / screening pipeline; best held-out test MAE 0.0702 eV.
Built a chip-surface physical-fingerprint verification workflow from BCP self-assembly random nanopatterns and SEM images.
Representative results: 126 SEM images; AUC 0.977; NanoPUF Research Platform.
Continued turning research processes into products: ExcitationNexus and the NanoPUF Research Platform, stressing data, computation, models, audit, and result tracking.
Representative results: from offline model experiments to reusable research workflows.
Advanced the Yuanju AI short-video mini program and ViolinSense AI (AI music × violin practice) prototypes.
Representative results: cloud Alpha and prototype validation; continued exploration of agent workflows and real usage scenarios.
Continuing AI for Science research at Shanghai Jiao Tong University's School of Chemistry and Chemical Engineering: molecular machine learning, quantum chemistry, research agents, and interdisciplinary problems.
Representative results: research, engineering, and product lines running in parallel.
PhD researcher in AI for Science: molecular machine learning, quantum chemistry, and research agents.
BEng, AI4E track; selected coursework: Advanced Mathematics I 96, Python Programming 100, Physical Chemistry 100, Colloid Chemistry 96, Chemical Engineering Principles II 95.
Deepening excited-state quantum chemistry, multi-objective prediction, uncertainty / OOD, and active screening, while exploring spin chemistry, CISS, and broader scientific intelligence. The goal is a stable personal research mainline and long-term, iterable research tooling.
From hydrogen-production optimisation to AI for Science — happy to talk.