My research brings together integrated photonics, nonlinear optics, and machine learning. The goal is to build optical hardware that can carry out computation directly in light. Matrix multiplication is the core piece, since it dominates the cost of running neural networks, but it is not the whole picture. How best to handle the other steps, such as the nonlinear activation, is still an open question that I am working through. Light is well suited to this work because it can carry many signals at once, at high speed, and with very little energy per operation.

To do this, I use a frequency comb as a light source. A comb provides many evenly spaced, coherent colours of light from a single device. By placing data on different colours, I can run many calculations through one photonic circuit at the same time, using colour as an extra dimension for parallel computing.

My work is shaped by a mixed background across quantum optics, astrophotonics, and quantum computing. I use this experience to connect the underlying physics of light with the practical demands of building real computing systems.