Roy Liu

Relevant Coursework: Data Structures, Discrete Mathematics, Statistics, Multivariable Calculus

• Developed a Bayesian inference framework using Physics-Informed SGLD to reconstruct 9D Hamiltonian parameters from experimental data.
• Architected a differentiable Modulated SIREN surrogate with 13 inputs, leveraging a global latent conditioning space to achieve ∼0.02 MSE.
• Engineered a high-throughput data generation pipeline in Julia on TACC supercomputers, producing 30GB of data.

• Modeled limited-slip differential torque behavior using custom differential-algebraic equations in Julia.
• Analyzed damping curve data and performed interpolation to optimize vehicle performance.
• Developed kinematics model features in Python/Matplotlib, including anti-roll bar angles.

• Engineered ML models (XGBoost, CNNs, SVMs) to predict stock movements with 86% accuracy.
• Integrated model with Alpaca API for real-time automated trading simulations.

• Ranked 3rd at USTA Zonal National Competition.
• Played 1st Singles, leading high school team to State Finals and County Championship.
• Developed resilience and time management skills through rigorous daily training.