I was a PhD student in the ECE department, working with Patricio A. Vela, in the Intelligent Visual Automation Lab. I graduated in January 2014. After a short stint at Oklahoma State University, where I worked in Distributed Autonomous Systems Lab with one of my principal collaborators Girish Chowdhary, I moved to Pindrop Security as a research scientist, where I work on developing machine learning algorithms and frameworks for audio fraud detection in an industrial setting.

Research Interests

I am interested in both the fundamental theory of machine learning, and its applications, particularly to control theory, and by extension, robotics.

In previous research, I applied theoretical work in sparse modeling to applications in machine learning and control theory, where 'sparse modeling' roughly means building constrained models on a budget from large amounts of data. These constraints tie in nicely with regularization theory, which tries to avoid overfitting on training data.

This work also demonstrates that a deeper understanding of machine learning techniques affords us the opportunity to solve difficult problems in nonlinear control. We have applied our techniques to the control of Unmanned Air Vehicles such as quadrotors, with future work targeting manipulators. For more, please see my dissertation. I have also worked on applications of machine learning to medical image processing, particularly skin cancer.

While my previous work utilized fundamental tools of machine learning to solve problems in control theory, my current focus is the opposite direction. The modern theory of machine learning has evolved with a focus on data that's static in time. There are many problems, particularly those arising from physical systems, where one must perform inference on time-varying functions, often for the use of decision-making and control. Control theory focuses on these problems, but lacks the nice functional tools of machine learning. My recent work is aimed at linking these two fields, in particular kernel machines and linear systems theory, which leads to methods that are theoretically well-founded, and practical. You can find out more on this on the Function Observers page.

My Google scholar page can be found here.


If you're interested in learning more about the work I did in my PhD, you can download my thesis here, and my dissertation presentation here.


My email is the first letter of my first name and my last name at gmail.com.