Research Interests

• Energy-efficient accelerators for physical AI models, including vision-language-action (VLA) models and World Action Models (WAMs), through algorithm-architecture co-design.
• Efficient chips and architectures for general-purpose models, including large language models (LLMs), diffusion models, and convolutional neural networks (CNNs), through algorithm-architecture co-design.
• In-memory-computing architectures and device co-design for efficient combinatorial optimization solvers, especially probabilistic Ising machines.