My research centers on developing Artificial-Intelligence (AI)-driven operando modelling frameworks for complex chemical systems in energy conversion and sustainable chemistry.

A major challenge in computational chemistry is that many experimentally relevant processes occur at time and length scales that are inaccessible to conventional electronic-structure calculations. My goal is to bridge this gap by integrating first-principles calculations, machine-learned interatomic potentials, active-learning strategies, and large-scale molecular simulations.

To address these challenges, I combine advanced electronic structure methods, molecular simulation, and machine-learning-assisted approaches. Unlike idealized static models, my research focuses on realistic operating environments, incorporating explicit solvent effects, finite-temperature dynamics, and constant-potential conditions. Through these approaches, I aim to uncover the fundamental roles of reaction microenvironments, potential-dependent charge transfer, and dynamic solid–liquid interfaces in governing chemical reactivity and materials performance.

Key Research Areas

Carbon Cycle Catalysts

Developing AI-driven operando modelling frameworks to understand and design transition-metal and single-atom catalysts for CO₂ conversion into fuels and value-added chemicals. By integrating machine-learned interatomic potentials, active-learning workflows, large-scale molecular simulations, and first-principles calculations, my research aims to reveal how dynamic reaction microenvironments, solvent effects, and interfacial phenomena govern C–C coupling and multi-carbon product formation.

Nitrogen Cycle Catalysts

Applying machine-learning-assisted multiscale simulations to investigate sustainable ammonia synthesis and nitrogen transformation processes. My work focuses on understanding N≡N bond activation under realistic operating conditions, integrating electronic structure theory, operando modelling, active-learning methodologies, and thermodynamic/kinetic analysis to accelerate the discovery of efficient non-precious catalysts.

Water Cycle Catalysts

Developing predictive operando simulation frameworks for water-splitting and fuel-cell electrocatalysts. Through the integration of machine-learned potentials, reactive molecular dynamics, active-learning approaches, and first-principles calculations, I aim to uncover the dynamic origins of activity, selectivity, and stability in OER and ORR systems, including catalyst reconstruction, spin-state effects, and support–catalyst interactions.

My computational toolkit spans electronic structure theory, machine-learning potentials, molecular simulation, and workflow automation, including VASP, CP2K, Gaussian, Materials Studio, MedeA, LAMMPS, GPUMD, MACE, PLUMED, and Python-based high-throughput workflows. These capabilities support scalable, reproducible, and data-driven catalyst discovery across diverse chemical environments and operating conditions.