Fourth-year student at Rice University
B.S. Operations Research | B.A. Mathematics
I am broadly interested in mathematical programming, online algorithms, and optimization under uncertainty.
I'm a fourth-year undergraduate at Rice University graduating in May 2026 with a B.S. in Operations Research and a B.A. in Mathematics. I spent Spring 2024 studying abroad at Budapest Semesters in Mathematics, where I deepened my understanding of discrete math and combinatorics.
My research focuses on interpretable algorithms and data-driven optimization. Inspired by my mentors, Sebastian Perez-Salazar and Emma Savannah Johnson, and my incredible team at Sandia National Labs, I am developing a framework called DIAL: Data-driven Interpretability of Algorithm Logic.
The goal of DIAL is to understand algorithms using polyhedral abstract interpretation. Once we have embedded or over-approximated an algorithm as a polytope, we can equip this polytope with an objective to optimize, resulting in a linear program (LP). The framework is named 'DIAL' because it allows us to systematically probe this LP. By "turning a dial," we can decide how much to interdict the optimization. This traces the full spectrum of performance—from the unconstrained optimum (the max solution) to the fully adversarial, worst-case outcome (the min-max solution). This process maps out a multi-dimensional response surface, where the algorithm's objective value is a direct function of the interdiction strategy.
This website contains several of my research ideas surrounding DIAL, including slides, a demo on ordinal threshold algorithms for k-secretary, and applications to neural network interpretability.
PDF slideshow summarizing some of my Sandia National Laboratories work.
This demo was created in collaboration with Charlie Cruz and Burton Alexander as the final project for my course COMP414 - Algorithms, Complexity, and Approximation.
The main directions for future research include: expanding this framework to substantially bigger neural networks, applying DIAL to trained neural networks using the polytope representation, and investigating whether this framework can be extended to more complex architectures including Recurrent Neural Networks (RNNs), Convolutional Neural Networks (CNNs), and most interestingly, Transformer models.
This is a personal (and incomplete) reading list that I'm hosting on my website for convenience and in case anyone is interested. Nodes represent papers and books, edges show literature I'm associating together, and the overall graph structure helps me organize my learning more effectively than randomly reading papers.
For several weeks, I worked on a small peonies farm in Homer, AK. We picked, cleaned, and packed the flowers for shipping. The living experience was really rustic. It was nice to live in nature, as it felt peaceful and restful.
My first time meeting "chess tourists" was in the Hungarian thermal bath house. They were visiting from Denmark and Moldova. My friend and I and three Kazakh college students were their challengers; we had a tournament lasting several hours!
My friend and I planned a trip to Eger. I assumed we were taking public transportation, but little did I know that we would cycle ~30 miles that day. During the cycling, we bought and drank a whole liter of grape juice.
I was visiting my friend at the University of Regensburg for his 21st birthday. After a night of sports pubs and debauchery, he treated us to a traditional sausage, pretzel, and coffee breakfast.
One of my favorite activities is polar plunging. It is a test of my discipline and feels so great after a dry sauna. This picture is of me in a village near Stockholm in 33 degree Fahrenheit water.
I enjoy reading many genres of books including philosophy, history, novels, and sci-fi.