Silicon and silicon nitride photonic integrated circuits (PICs) are taking on increasingly demanding roles in optical communications, sensing, and quantum applications, yet device count and functionality continue to outpace the area budget available on a chip. Fitting more capability into a fixed footprint is now as much a design problem as a fabrication one — a class of problem where modern machine learning, including recent generative AI methods, has shown promising results compared to conventional design approaches.

In this talk, I survey recent progress in applying ML and inverse design to photonic device development, drawing on work from across the community as well as projects I have led. I then present our recent work on compact silicon nitride waveguide bends, in which AI-assisted design achieves lower loss than a widely-used Euler bend of similar footprint. We have also generated arbitrary-angle bends for flexible routing. The designs have been verified through foundry fabrication and measurement, indicating that the approach is compatible with standard production flows rather than confined to simulation. Looking ahead, AI and ML may help accelerate process design kit (PDK) development by supporting component library generation.

This is collaborative work with Spark Photonics Design, Inc. and AIM Photonics.

Keisuke Kojima has worked in photonic devices and optical communications for over 40 years. He received his B.S., M.S., and Ph.D. from the University of Tokyo, Japan, and an additional M.S. from the University of California, Berkeley. He spent eight years at Mitsubishi Electric in Japan working on narrow-linewidth/surface-emitting DFB and DBR lasers and optical neural networks, followed by nine years at AT&T/Lucent Bell Laboratories on the development and commercial introduction of uncooled Fabry–Pérot and DFB lasers, and research on VCSELs. From 2005 to 2021 he was with Mitsubishi Electric Research Laboratories (MERL) in Cambridge, MA, where he focused on nanophotonic device design using deep learning and coherent optical systems. In 2022 he founded Boston Quantum Photonics, dedicated to developing novel photonic devices using state-of-the-art AI/machine learning. He is also a Senior Photonic Device Designer at Cactus Materials, Inc. in Tempe, AZ, where he designs semiconductor lasers. He is a Fellow of Optica and a Life Fellow of IEEE, and served as Chair of the IEEE Photonics Society’s Boston Chapter.

Register for this event.