While in graduate school, my research focus was in Topological Data Analysis, and specifically in addressing aspects of the field that make it unapproachable by outsiders.
My efforts have been geared towards expanding usability and interpretability of existing methods, developing easy to use tools to make the existing techniques more accessible, and exploring new applications to expand the possible use cases of TDA.
Since leaving graduate school, I have been focused on machine learning engineering, organizational theory as it relates to development of intelligent systems within companies, software architecture or machine learning systems, and serverless machine learning. I have found that the problems related to people, product development, and machine learning design are often more challenging and engaging than the technical problems of prediction and inference.
In support of my research and many others, I develop and maintain the Scikit-TDA Python package. This has been a monumental undertaken and has completely consumed my research and development time. My role has ranged from implementing and incorporating new algorithms to developing documentation and build systems to managing social media and promotion.
This project grew from an implementation of Mapper to an all encompassing system for nearly all common methods used in Topological Data Analysis. The project has had overwhelmingly positive support from the TDA community and has a growing list of contributors.
You can learn more from my presentation at Scipy 2019, or at github.com/scikit-tda. We are always looking for more people to help maintain and improve the libraries.
While visiting Pacific Northwest National Laboratory, I collaborated with Dustin L. Arendt to develop new ways Mapper could be applied for explaining machine learning models.
Initial progress in this direction will be a part of the IEEE VIS 2018 Workshop on VISxAI. You can read more about the work here.
In collaboration with Bei Wang and Bala Krishnamoorthy, I am developing ways of composing Mappers built from different filter functions together, and developing techniques for quantifying the topological correlation and complexity inherent in particular filter functions.
I presented preliminary results of this work at the Young Researchers Forum at SoCG 2018 and at SIAM CSE 2019. You can find the abstract here.