• Accelerated large-scale simulations by 5.6x through efficient utilization of GPU cores for massive agent parallelization, resulting in a significant boost in library performance.• Created two benchmark applications, a traffic simulation and a tuberculosis simulation, to validate the accuracy and evaluate the performance of the MASS CUDA library against other agent-based libraries.• Presented research findings at a university conference, attended by 100+ students, researchers, and PhDs, which resulted in continued funding for the development of the MASS CUDA library.• Spearheaded an initiative that improved functionality and usability of the MASS CUDA library by 25% through identifying and resolving 6 major bugs.• Boosted application performance by 12% through the implementation of a direction-based collision synchronization algorithm.