Physics PhD candidate at HKUST, building reliable AI agents, research evaluation pipelines, and privacy-aware tool systems.
I am interested in systems where AI agents can execute long-horizon technical work while humans retain control over objectives, evidence, permissions, and risk.
- Scientific and research-agent evaluation
- Reproducible empirical ML workflows
- Tool-use security, privacy, and auditability
- Experiment automation and evidence-grounded reasoning
My core PhD research: measuring piconewton-scale quantum-fluctuation forces with a custom MEMS platform. I integrate semiconductor nanofabrication, materials characterization, low-noise lock-in instrumentation, experiment automation, and reproducible analysis to turn weak physical signals into auditable scientific evidence.
A privacy-aware payment control plane for agents purchasing x402 services. The project uses constrained budgets, disposable payers, encrypted recovery records, and auditable receipts. It has been validated with real paid API calls on Base using USDC.
My physics research focuses on high-vacuum precision measurement and weak-signal detection. I have built experimental systems involving optical readout, lock-in detection, noise analysis, and LabVIEW automation.
I also work on leakage-safe empirical evaluation, including temporal validation, embargoed splits, block bootstrap, and the distinction between predictive evidence and deployable performance.
Open to research engineering, scientific AI, ML evaluation, quantitative research, and research-automation opportunities.