I am a Machine Learning practitioner drawn to the places where philosophy, physics, and mathematics touch. The same impulse that compels a physicist to reduce a falling body to a differential equation compels me to reduce a messy real-world signal — a breathing chest, a price series, a language model's behaviour — into something clean, testable, and defensible.

My work spans four directions that I refuse to keep separate. A contactless respiratory digital twin that fuses Wi-Fi Channel State Information with a Physics-Informed Neural Network constrained by the single-compartment lung equation. An event-driven C++ backtesting engine that makes lookahead bias structurally impossible, then quantifies how much apparent alpha dissolves under realistic costs. A design-science security protocol — VeriData — that unifies AI authenticity verification, selective encryption, and blockchain-anchored integrity for post-quantum environments. And AI safety & compliance frameworks grounded in NIST AI RMF and the EU AI Act.

VeriData is the project I am most fond of, because it did not arrive from the literature — it arrived as a raw intuition that the same pattern (verify, then protect, then anchor) was missing from three separate domains at once. Only later was it validated into a formal five-phase protocol through design science research. That is the shape of work I want to spend my life doing: intuition first, rigor second, and a refusal to believe that disciplines must stay in their lanes.

I am most at home when a problem requires three things at once: a mathematical model, a physical intuition, and a philosophical honesty about what the model cannot tell me.