Energy-Efficient
AI Processor

During my undergraduate studies, I conducted research on low-power digital AI processors at Kuroda Lab, University of Tokyo.
As a result of my thesis, I taped out a chip measuring 3mm by 4.5mm using a TSMC 40nm process.
I personally led all stages of the project, from proposing the algorithm and implementing it in PyTorch to overseeing the final chip fabrication.
To the left is a photo of the actual chip that I taped out.

Optical Neural Networks
using Silicon Photonics

From graduate school, I changed my field to Silicon Photonics at Takenaka Lab, University of Tokyo.
I believe that using light instead of electricity not only in communication but also in computation could achieve ultimate low power consumption, which is what I am aiming for.
I am currently advancing research toward the demonstration of a high-performance optical phase shifter using novel materials, as well as a tensor core that utilizes it.
(The image on the right is just a free image of an optical fiber.)