Uncertainty-Aware Systems That Infer, Decide, and Act

Research in system optimization, control, mechanistic machine learning, adaptive sensing, and closed-loop experimentation.

Research Program Methods

I am a researcher and engineer working on methods for systems that must learn from uncertain measurements and choose useful next actions. My current direction is mechanistically grounded machine learning for adaptive biomedical sensing and closed-loop experimentation.

The red thread from my earlier work is not a single application domain. It is the design of models, estimators, optimization methods, and control strategies for real systems operating under uncertainty.

About / Trajectory

• 2026–present: Researcher, Gongshu Gongda Future Technology Research Institute (Zhejiang University of Technology), Hangzhou.
• 2026–present: Guest Researcher, Hangzhou Institute of Technology Xidian, Xidian University — adaptive biomedical sensing and closed-loop experimentation programs (part-time).
• 2026–present: Foundation Algorithm Library Developer, HRK-Data (part-time).
• 2020–2024: Postdoctoral Researcher, ETH Zurich — energy systems, automation, modeling, and control. Concurrent Scientific Officer and Data Manager, NCCR Automation.
• 2023–2024: Lecturer and laboratory lead, ETH Zurich — Quad-Rotors and Control Experiments Lab (~120 + ~200 students).
• 2016–2020: PhD Researcher, DTU Compute — efficient operation, modeling, optimization, and control of energy grids.

Full biography and CV →

Research Focus

Uncertainty-aware inference

I work on probabilistic reconstruction and state estimation methods that combine measurement data with physical or scientific structure, producing uncertainty estimates that can support decisions.

Adaptive measurement and experiment selection

I frame sensing and experimentation as sequential decision-making problems: what should be measured, sampled, tested, or stopped next, given current uncertainty and practical constraints?

Closed-loop experimental systems

My current research connects sensing, inference, and action in systems that can adapt their measurement or experimental strategy as evidence accumulates. Biological experimental actions remain expert-reviewed; safe sensing-system actions can be automated under constraints.

Mechanistic machine learning

Learned representations are constrained by physical forward models, scientific priors, and domain knowledge, so predictions remain interpretable, extrapolate credibly, and support actionable decisions.

Current Platforms

Biomedical sensing is the primary adaptive sensing direction, with radar-based physiological monitoring as a testbed for physical measurement models, inverse problems, real-time constraints, and health-relevant signals.

Closed-loop discovery is represented by AntiSyn-AI and related biomedical collaborations (manuscript under review), where uncertainty-aware active selection can help prioritize informative combinations, concentrations, or experimental conditions.

Selected Contributions

These outputs demonstrate the methodological base across application domains:

Two additional manuscripts are under review. See the publications page for the complete record.

Prior Foundations

My earlier work at DTU and ETH Zurich developed methods for microgrid control, grey-box modeling, energy-system optimization, state estimation, and automation teaching. These projects remain important evidence, but on this site they are framed as foundations for a broader research program in uncertainty-aware modeling, optimization, and control.

Collaboration

I welcome inquiries about research collaboration, adaptive sensing and experimentation, uncertainty-aware modeling, and related academic opportunities. Consulting engagements in machine learning, optimization, and control are considered on a project basis.

Contact → · View Publications →