Research

My research sits at the intersection of machine learning and security. My dissertation emphasizes AI-based malware detection, analysis, and mitigation, but I hope to expand out into other areas, particularly agentic security. I did some other things before I got into security, so here is a high-level overview of my trajectory.

Machine Learning Security @RIT (2022 - 2027)

My Ph.D. research, advised by Dr. Matthew Wright and Dr. Yin Pan, has focused on end-to-end malware detection: deep learning models that operate directly on executable binaries rather than on features engineered by analysts. My work tackles this problem from several angles, including representation learning, architecture design, and scaling laws. Before settling on this direction, I explored adversarial malware generation, methods for explaining malware classifiers’ decisions, and, for a few months, network traffic analysis. I’ve been fortunate to collaborate with a number of researchers from academia (Ariel University, University College Cork, Queen’s University Belfast) and industry (Google Cloud).

Digital Authenticity @CACI (2021)

In the summer of 2021, I interned in CACI International’s cybersecurity division, where I worked on telling real content from fake. My first project used neural networks to detect deepfake audio, which was a fairly new problem at the time and has since grown into a major field. In the second, I designed feature sets for machine learning to flag manipulated PDF documents. This was my first exposure to security, and it’s what hooked me on the field.

Learning Curve @TCNJ (2021 - 2022)

Next, I joined Dr. Michael Bloodgood’s Learning Curve group, which works on building natural language processing systems with less labeled data. Labeling is often the most expensive part of machine learning, and active learning cuts the cost by letting the model choose which examples are worth labeling. The catch is knowing when to stop: stop too early and the model underperforms, stop too late and you’ve wasted the effort. I showed that the data set aside for making that call can significantly change the outcome, something recent work had mostly overlooked. That became my first research paper, which was nominated for Best Paper at IEEE ICSC 2022, and seeing a project through from idea to publication is what convinced me to pursue a Ph.D.

Collective Intelligence @TCNJ (2020 - 2021)

My first computer science research was in Dr. Monisha Pulimood’s Collective Intelligence group, which, with support from the NSF, was building a web application called CABPortal. Academic search engines tend to favor highly cited papers from prestigious venues, and CABPortal was meant to help researchers, especially those working across disciplines, find the relevant work those engines bury. I built its machine learning recommendation system, and because the platform was brand new and had no user activity to learn from, much of my time went into figuring out how to train and evaluate a recommender without it. That project is what got me interested in machine learning.

Photonics @TCNJ (2018 - 2020)

I started college as a physics and computer science double major and joined Dr. David McGee’s photonics group, which develops new ways to fabricate optical microstructures (the tiny surface patterns used in diffractive optics and holograms) by writing them into light-sensitive polymer films with polarized laser light. My job was writing the software that ran the lab’s hardware, like its lasers and motors, along with an application for running fabrication experiments. I worked closely with researchers from the Technical University of Berlin, and my software was used to run the core experiments in three co-authored papers. The bigger takeaway for me was how much I loved programming (C++ excepted), so I dropped physics to a minor and focused on computer science.