Home

Expertise

Research

Publication

On-device AI and Intelligent Sensors towards biomedical, robotics, and advanced AI computing hardware.

"Computing Where Data Is Born" — this is the vision driving my research.

Today's AI is limited less by how fast we compute than by how far data must travel — from sensor to processor, from memory to logic.
We build intelligent sensors that encode and analyze signals the instant they are generated, and AI hardware that fuses analog, digital, and photonic computing. This allows processing information in massively parallel form and route it in three dimensions across electronic–photonic engines. Through heterogeneous integration, intelligence lives inside the hardware itself.

Research Topic

Robotics and Smart Health

On-device AI using Analog/Digital signal computing

Advanced Semiconductor Growth and 2D Materials

Heterointegration and Layer Transfer Technologies

Wearable Multimodal Sensors

Recent works

Schematic of In-sensor and near-sensor computing

In-sensor and Near-sensor Computing

Artificial intelligence technology transforms traditional sensors from passive data collectors into active computing nodes, performing data processing at the edge. This paradigm shift toward in- and near-sensor computing mitigates inherent inefficiencies associated with data traversal between sensing, memory, and processing units.

Schematic of In-sensor and near-sensor computing

In-sensor and Near-sensor Computing

Artificial intelligence technology transforms traditional sensors from passive data collectors into active computing nodes, performing data processing at the edge. This paradigm shift toward in- and near-sensor computing mitigates inherent inefficiencies associated with data traversal between sensing, memory, and processing units.

Schematic of In-sensor and near-sensor computing

In-sensor and Near-sensor Computing

Artificial intelligence technology transforms traditional sensors from passive data collectors into active computing nodes, performing data processing at the edge. This paradigm shift toward in- and near-sensor computing mitigates inherent inefficiencies associated with data traversal between sensing, memory, and processing units.