Overview

My research lies at the intersection of materials science, mathematical modeling, numerical methods, and scientific computing. I work on computational approaches for understanding complex physical and engineering systems, with a particular focus on problems where the underlying structure or parameters cannot be observed directly and must instead be inferred from indirect, noisy, or incomplete information.

A significant part of my research has focused on microstructure and microtexture evolution in polycrystalline materials. During my PhD, I developed physics-informed stochastic and mechanistic models for nucleation and growth, along with computational methods for generating, analyzing, and reconstructing three-dimensional microstructures. This work resulted in EvoSim, a generic simulation framework for stochastic nucleation-growth processes, and REvoSim, an inverse modeling approach for reconstructing microstructures from limited grain-level information such as volumes and centroids.

More broadly, my work involves building mathematical models of physical processes and solving the computational problems that arise from them. This has included coupled thermo-mechanical modeling of hot rolling, dynamic recrystallization and grain-size prediction, finite-element modeling of carburizing and quenching, phase-transformation kinetics, stochastic parameter estimation from experimental data, and computational analysis of microscopy images.

A recurring theme in my research is the use of computation as a way of performing virtual experiments: formulate the governing relationships, explore the behavior of a system numerically, infer hidden parameters or structures from available observations, and develop algorithms that can reproduce or predict the physical behavior. This often combines mechanistic models with stochastic methods, optimization, numerical computation, image analysis, and statistical techniques.

I am particularly interested in extending this computational approach toward data-driven methods and machine learning, using them alongside physics-based modeling rather than treating them as a replacement for physical understanding.