Biographical Information:

I am a senior level doctoral student in the GRASP lab at University of Pennsylvania. I work with Ali Jadbabaie in the Department of Electrical and Systems Engineering.

I have a Masters of Science in Electrical Engineering from the University of Pennsylvania, a Bachelors of Engineering from the Thayer School of Engineering with a focus in control theory and a Bachelors of Arts in Engineering Science from Dartmouth College with a focus on modeling of environmental systems: honors thesis on Adoption Dynamics of Green Technology.

My research interests includes: Network Science, Convex Optimization and Analysis, Game Theory, Network Optimization and Control, and Distributed Optimization Methods. I am interested in applications to wired and wireless networks and sensor networks, as well as economic models, marketing and social network applications.

My thesis topic is Distributed Convex Network Flow Optimization. I have developed Accelerated Dual Descent (ADD) an approximate Newton's method and generalized it to the case of packet routing problems with Accelerated Backpressure (ABP). I have worked closely with Alejandro Ribeiro. See my publications.

I have also worked on Network Science tools for generalized graphs (simplicial complexes). We rank edges in a simplicial complex based on a generalized Page Rank operator. Hodge Decomposition of the rankings tell us whether the edge is important due to holes, sparse cuts or cliques. Some results of these tools can be found here:
Dumbbell Demo
Small Proximity Network
Large Proximity Network
Collaboration Network
Collaboration Network Generated by Publication Date
The first four videos show variation over the parameter beta (encodes teleportation probability) and in the last the complex changes with time. The edge thicknesses indicate their "edge page rank" and the color indicates which Hodge subspace contributes primarily to its rank.

My most recent work is on contagion dynamics in network systems with Victor Preciado. We control prevention and recovery resources as well consider interaction limitations to drive the contagions out of the system of interest. Our optimization framework includes semidefinite and geometric programs.