Research
Four interlocking threads, one mission: sustainable computing for the AI era.
Our work spans the full stack of emerging computing: from device physics and analog photonic circuit design, to system architectures and warehouse-scale photonic interconnects, to the ML algorithms that design these systems in the first place. Each thread sharpens the others.
Research network
The four areas
Photonic Computing & Architecture
Designing the chip and the system that runs on light.
We work at three levels — device physics, photonic circuit and architecture design, and full-system simulation. We also design the mixed-signal interface circuits (drivers, TIAs, ADCs/DACs) that bridge analog photonic cores to digital compute hierarchies.
Read more → Area 02Photonic Data Center Networks
Wiring up thousands of accelerators with light.
Training a frontier AI model is bottlenecked by communication, not compute. We design photonic interconnects and warehouse-scale topologies for distributed training and inference, and ask how the workload should be re-shaped when the physical layer changes.
Read more → Area 03Inverse Design for Photonic Devices
Letting algorithms design our devices — and using GPUs to do it fast.
Specify the desired optical response, let the optimizer find the geometry. We accelerate inverse design with GPUs and ML surrogates, scaling the workflow from single components up to subsystem-level simulation.
Read more → Area 04Hybrid FPGA Integration
Bringing photonics and memristors to existing reconfigurable hardware.
ASIC photonic chips are years out; FPGAs are everywhere. We integrate silicon photonic interconnects and non-volatile memristive memories with FPGA platforms to build hybrid accelerators that are deployable today.
Read more →