This project develops the modelling methods for personalized brain models using the Human Connectome Project (HCP) cohort, an open source datasets. The work has two parts. First, we process the diffusion MRI of 77 HCP subjects through fibre-orientation modelling and anatomically-constrained probabilistic tractography, producing per-subject and group structural connectivity and fibre-length matrices. Second, using automatic differentiation (TVB-Optim, JAX), we estimate the regional parameters and the structural connectivity of a whole-brain (Hopf) model by gradient descent against each subject's functional connectivity, its temporal dynamics and its power spectra — both on the group mean connectome and independently for each of the 77 subjects.