My research lies at the active intersection of theoretical condensed matter physics, theoretical machine learning, and quantum information. I specialize in non-equilibrium quantum transport theory, high-dimensional probability for neural scaling laws, symbolic quantum mechanics, and machine learning methods for physical systems.
McGill Department of Mathematics and Statistics • Supervised by Prof. Elliot Paquette & Prof. Courtney Paquette
Conducting research in theoretical machine learning focusing on Power Law Random Feature models. We leverage toolkits from random matrix theory on Volterra equations, high-dimensional probability, and optimization theory to mathematically derive precise scaling laws for Large Language Models (LLMs).
A central question of our work is investigating how anisotropy builds up within the model during training and how this anisotropic feature space directly influences learning dynamics and generalization bounds.
McGill Physics Department • Supervised by Prof. Hong Guo
My doctoral research centers on theoretical and computational physics with the Keldysh non-equilibrium Green function (NEGF) formalism. Key thrusts include:
SymGF): Developing and extending our in-house package that symbolically expands quantum transport equations from the Bogoliubov–Born–Green–Kirkwood–Yvon (BBGKY) hierarchy using the Heisenberg equation of motion. We combine LLM symbolic reasoning to simplify self-energy series for transport in strongly correlated qubit arrays.
Kastler Brossel Laboratory (Sorbonne Université, ENS, Collège de France) • Supervised by Prof. Quentin Glorieux
Conducted research with the Quantum Fluid of Light Group in Paris. Designed and trained deep machine learning architectures using theoretical simulations of atomic optical setups. Successfully solved the inverse problem to recover experimental parameters from single-shot optical measurements of the non-linear Schrödinger equation in hot atomic vapors.
McGill Physics Department (Quantum Nano Electronics Laboratory - QNEL) • Supervised by Prof. Michael Hilke
Designed a novel control protocol to investigate decoherence dynamics of qubits coupled to a 1D Su-Schrieffer-Heeger (SSH) topological chain. Demonstrated that topological edge states can robustly protect qubit coherence even under significant environmental coupling.
INRIA (Institut national de recherche en informatique et en automatique), Nancy, France • Supervised by Dr. Marie-Dominique Devignes
Worked within the CAPSID team (Computational Algorithms for Protein Structures and Interactions), applying recursive combinatorial algorithms to generate, evaluate, and filter candidate molecular structures.
Banner photo by Freepik