My research develops reproducible computational frameworks integrating first-principles physics, machine learning, and near-term quantum algorithms to address bottlenecks in materials discovery and chemical simulation.
I develop and apply quantum subspace and spectral filtering techniques (VQE, qEOM, SQD, and filtered-QPE) to resolve low-lying eigenstates, excitation energies, and dissociation pathways in molecular and materials systems where classical methods face exponential scaling.
My focus is on algorithmic structure—active-space design, spectral isolation, and noise-aware subspace compression—rather than hardware benchmarking or black-box demonstrations.
I build machine-learning models for materials property prediction and screening, with an emphasis on voltage prediction for alkali-ion battery materials. My work integrates physics-informed descriptors derived from DFT with modern neural architectures to improve accuracy, robustness, and interpretability.
Beyond predictive performance, I analyze feature importance, uncertainty, and data bias to ensure that ML models provide scientifically meaningful guidance rather than purely empirical correlations.
I design high-throughput DFT and AIMD workflows for structural relaxation, thermodynamic analysis, and data extraction, with a strong emphasis on automation, provenance, and reproducibility.
These workflows serve both as standalone computational studies and as structured data generators for downstream machine-learning and quantum simulation pipelines.
I apply quantum chemistry methods to molecular systems relevant to energy storage and functional materials, systematically benchmarking quantum algorithms against classical electronic-structure calculations.
This work aims to identify realistic problem classes where near-term quantum methods provide qualitative or quantitative advantages, while clearly delineating their current limitations.
