The computational burden of iterative Born methods is immense. Therefore obtaining real-life images is either impossible or impractical with conventional computational approaches, and previous research on Born methods was limited in terms of real-life applications. As a remedy, our research effort makes use of fast algorithms and parallel supercomputing for solving large and realistic multiple-scattering problems for obtaining physically meaningful images. There are two parallelization challenges here: the first one is the parallelization of independent scattering solutions, and the second one is the parallelization of forward-scattering solver itself. We tackle the both challenges. This report focuses on our philosophy for efficient utilization of frameworks with hundreds of computing nodes with heterogeneous CPU+GPU architectures.