Ducted fans have found renewed interest particularly in the field of micro aerial vehicles. The complex flow interactions between the rotor, shroud and nacelle hinder the design of optimal lifting duct shapes. Further, the optimum duct shape changes with freestream velocity and rotor configuration. Analytic equations are insufficient to model and predict the performance of ducted fans, yet general CFD solvers are too slow to be used for optimization. For this research, a genetic algorithm is coupled with a simplified CFD solver based on a momentum source rotor model, creating an efficient shape optimization technique for ducted fan systems. The performance gains of the genetic algorithm are documented and the optimum duct shape for several configurations are presented as proof of concept.