I conduct empirical research in mechanistic interpretability and sparse autoencoders (SAEs), decomposing neural representations into monosemantic feature dictionaries and topological manifold trajectories.
Beyond deep learning, I study aerodynamics as an FAA Student Pilot (Private Pilot expected Summer 2026), perform as a trumpet section leader in the Raiders Drum & Bugle Corps, and direct student hackathons.
Empirical work isolating deceptive reasoning sub-circuits, analyzing high-dimensional latent space manifolds, and benchmarking model evaluation scorecards.
Integrating physical flight training towards the FAA Private Pilot Certificate with low-level computational fluid dynamics engines written in C and OpenGL shaders.
Solving incompressible Navier-Stokes momentum equations to model boundary layer separation, stall vortex dynamics, and vortex-induced drag over cambered airfoils in real-time GPU simulations.
Open-source mechanistic interpretability repositories, autonomous agents, and systems engines.
Touring brass ensemble leadership, aviation flight hours, software quality engineering, and national hackathon organization.