Hydrogen embrittlement (HE) poses a significant challenge in utilizing hydrogen as a clean energy carrier, particularly in steel. This project focuses on exploring how the microstructure of iron or steel, especially grain boundaries and dislocations, affects hydrogen diffusion and aims to elucidate the atomistic mechanisms underlying HE. Traditional methods of experimentation and simulation face challenges with either accuracy or scalability. To address these issues, we will leverage AI-based, machine-learned interatomic potentials to accurately model hydrogen diffusion within atomic environments that mimic real microstructural features. Through dynamic simulations, we will measure the diffusivity of hydrogen across these environments and correlate these findings with experimental data. This approach aims to enhance our understanding of the microstructural impact on hydrogen diffusion, leading to the potential for reverse-engineering microstructures. Such advancements have broad implications, extending beyond HE mitigation to influence research and industrial applications where the role of hydrogen as an energy carrier is critical.