I am an independent researcher working at the intersection of machine learning, inverse problems, and mathematical physics. My current work focuses on understanding how learning-based methods can approximate or accelerate the solution of classical inverse problems, particularly in settings governed by differential equations and physical laws.

In 2025, my solo-authored paper “Deceptron: Learned Local Inverses for Fast and Stable Physics Inversion” was accepted for presentation at the NeurIPS 2025 Machine Learning for Physical Sciences workshop, where I study how local learned inverse operators can emulate Gauss-Newton-style updates without explicit Jacobians. I am now developing a deeper theoretical program on inverse problems.

Beyond my current projects, I aim to contribute to the foundations of scientific machine learning, bridging operator theory, optimization, physics, and deep learning. I am particularly interested in how high-dimensional inference behaves under structural constraints, and how learning-based algorithms can reveal new mathematical viewpoints on classical physical systems.

I am currently a Grade-11 science student in Surat, India, and plan to pursue advanced study and research in mathematics, physics, and machine learning at the university level.