Monte Carlo methods represent one of the most accurate approaches for modeling carrier transport in semiconductors, but their computational cost, particularly in terms of memory usage, often limits their applicability to large-scale or full-band simulations. This thesis presents a machine-learning enhancement of Monte Carlo simulations for electron transport in semiconductors, developed within an object-oriented Python framework. First, a fully object-oriented Monte Carlo code is implemented and validated for bulk GaAs, reproducing well-established transport properties such as velocity–field characteristics, valley occupancy, energy convergence, and autocorrelation functions. The modular structure improves flexibility, readability, and extensibility compared to traditional procedural implementations. Neural networks are then introduced to fit both analytical and numerical scattering rate functions.