Machine learning, computer architecture, systems, and circuits are among my research interests. The majority of my recent graduate and postdoctoral work has involved algorithm-hardware co-design, from algorithm optimization to chip design. For example, the most recent chip I worked on at Harvard Architecture, Circuits, and Compiler group was an Arm-based system-on-chip with multiple machine learning cores for various algorithms such as attention-based NLP and Bayesian inference. We demonstrated real-time speech recognition by compiling and running parts of the workload on the appropriate hardware.
Prior to my PhD, I worked at Samsung designing commercial processors for smartphones. I also worked at Qualcomm and IBM, where I developed efficient machine learning frameworks and architectures.
These days I’m interested in building ML-powered cloud computing system that one can trust to provide the best performance and how to make large language models and text-to-(image/video/anything) models more efficient.
Some things about me: