(385ab) Advancing Computational Materials Design By Leveraging Data-Driven Approaches, Multiscale Modeling, and Uncertainty-Aware AI

Research Interests

A key pillar of my work is the research methodology for developing Bayesian Uncertainty Quantification (BUQ), which I have applied to molecular simulations and force field development to assess predictive confidence and enhance decision-making across Technology Readiness Levels (TRLs). At Virginia Tech, I led efforts to couple BUQ (Ensemble MCMC sampling) with molecular simulations to improve MD model robustness and applied a novel PSO-integrated GA to optimize and design functionalized MOFs for hydrogen storage and CO2 capture. At Georgia Tech, under Prof. Rampi Ramprasad, my postdoctoral work centers on developing machine-learned interatomic potentials (MLIPs) for polymers using DeepMD, Allegro, and SO3LR, while also integrating large language models (e.g., GPT-4, LLaMA) with structure–property databases to streamline informatics-driven screening for industrially relevant polymers. Overall, these projects formed part of a larger portfolio of over 19 independent and collaborative research efforts I’ve led in the past 7 years, many of which are now published in high-impact journals. Throughout my academic career, I have worked in cross-collaboration, with chemists, materials scientists, and researchers from other institutions and U.S. national laboratories leading to impactful advances in hybrid materials, drug-delivery MOFs, and confined fluid dynamics.

I have also mentored 5 Ph.D. students and 4 undergraduates during my Ph.D. and postdoctoral tenure, assisting them in developing workflows and writing research articles. My commitment to scientific communication is reflected in 20+ conference presentations across 14+ national and international meetings, where I’ve presented complex and technical ideas clearly to diverse audiences. With hands-on expertise in classical MD and quantum DFT simulations (LAMMPS, NAMD, VASP), Grand Canonical Monte Carlo (RASPA 2.0), machine learning (TensorFlow, PyTorch, JAX), and large language models including in-context learning, fine-tuning, and retrieval-augmented generation (RAG) pipeline development, as well as high-performance computing, I aim to bring together scalable AI tools and physics-based modeling to deliver faster, smarter, and more reliable solutions in product development. Whether improving polymer electrolyte design or screening membranes for gas separation, my research is grounded in methodology/platform development that combines technical clarity, and practical value, which is critical for transitioning innovation from digital labs to industrial deployment.

Selected Publications (Please see Google Scholar for all publications):

• Photoactivated Multivariate Metal–Organic Frameworks for On-Demand Drug Release: The Role of Host–Guest Interactions

Journal of American Chemical Society (JACS), 2025

• Evaluation of Sampling Algorithms Used for Bayesian Uncertainty Quantification of Molecular Dynamics Force Fields

Journal of Chemical Theory and Computation, 2024

• Data driven discovery of MOFs for hydrogen gas adsorption

Journal of Chemical Theory and Computation, 2023

• Dual-Force Zone Nonequilibrium Molecular Dynamics Simulations on Nanoporous Metal-Organic Framework Membranes for Separation of H2/CH4 Mixtures

ACS Applied Nanomaterials, 2022

• Modelling drug adsorption in metal-organic frameworks: the role of solvent

RSC Advances, 2021