Q&A with Katelyn Watts

What project are you working on?

My work is focused on applying machine learning techniques to photonics problems. As part of a larger project on inverse design of subwavelength gratings, I investigate active learning methods with the goal of building a model that accurately predicts the relationship between grating geometry and effective refractive index. I am excited to be bringing this computational knowledge to physical systems in my active photonics coursework.

What problem(s) are you aiming to address through your research?

Each datapoint in the training set requires both a full 3-dimensional FDTD simulation and a 2-dimensional one. This means that to determine the complex relationship between grating geometry and effective index, we need to run tens of thousands of simulations. My work aims to significantly reduce that number while maintaining a high level of accuracy in the model’s predictions. More broadly, the project will create a suite of grating coupler designs that correspond to a range of refractive indices, providing photonic systems researchers with customizable components that better suit their needs.

What inspired you to pursue this program of study?

The NUCLEUS program gives me an opportunity to learn about photonics from a new lens. I’m excited to be working on physical systems and learning about experimental techniques while continuing to explore computational concepts. I’m looking forward to learning more about neuromorphic computing, especially from a quantum photonics perspective. The program also places emphasis on science communication, an essential skill for researchers that I am always looking to refine.