I am currently working on a project studying the mechanistic interpretability of a neural network trained to predict the properties of nuclei. In the past, I have worked on the real time analysis system (trigger) of LHCb which processes over 5 TB/s using GPUs and machine learning. My major projects have focused on the dark photon search, creating online monitoring of the trigger system as a whole, and implementing neural networks to improve particle identification.

I worked in Dr. Peter Mucha's lab using machine learning techniques to study a wide variety of data. My first project regarded voter registration rates in the state of North Carolina. Later I investigated the perception of alcohol use in collegiate social groups.

I worked with data from the ABRACADABRA detector in search of axion dark matter. I have completed studies to discover sources of noise in the data. I have also simulated various aspects of the experiment using COMSOL Multiphysics. I completed my thesis on the implementation of a positive feedback circuit using simulations and Verilog to program the filter on an FPGA.

I ported the Geant4 Monte Carlo simulation for a proposed experiment called KLEVER which will look for an extremely rare kaon decay. After the simulation was executable, I ran sensitivity studies and compared them with the less rigorous simulation that was previously used. This was in C++.

I worked with the NA62 experiment to look for an axion like particle. My project was to analyze the data taken over three years to find events matching the profile of the decay of neutral mesons into two photons. I used C++ and ROOT for this work. I was able to find an irregularity in the data and systematically save physics data that the collaboration was planning to discard. I also wrote an algorithm to mask noise from the calorimeter.

OpenSpace is software that visualizes the entire observed universe using real data. Spacecraft missions are one view in OpenSpace; each mission includes an accurate model of the spacecraft, the path it followed, onboard instrument timing, and data collected. I created visualizations of the Juno, Dawn, Mars Atmosphere and Volatile Evolution mission (MAVEN), Lunar Reconnaissance Orbiter (LRO), Stardust, and Deep Impact missions using SPICE kernels from the Navigation and Ancillary Information Facility and the Lua programming language. I developed a new method for the trails of the Juno, MAVEN, and LRO missions to accommodate for the separate traveling and orbiting phases. Images taken by Juno and Dawn are used to create textures on Jupiter, Ceres, and Vesta. To do this, I wrote Python scripts to download the images and create metadata files for each in the format OpenSpace reads.

During the summers and on weekends, I worked as a teller in the main branch of the credit union. This involved cashing checks, processing loan payments, and dispersing cashier's checks, among other things. During the school day, I worked at a branch within my school. My senior year, I worked alone at the school branch.

I analyzed data from the Mars Atmosphere and Volatile Evolution mission (MAVEN) spacecraft to learn about ion escape from the Martian atmosphere. The final project that I completed looked at the direction of the magnetic field at various altitudes, latitudes, and local times and compared those findings with data from the MGS mission. I also spent time working with data related to ion density, temperature, and velocity in the Martian atmosphere. I learned and used the IDL computing language.