An efficient and practical genetic algorithm (GA) was developed to optimize the UO2/Gd2O3 fuel pin burnable poison (BP) configurations for fresh fuel assembly (FA) designs loaded in a pressurized water reactor core. The objective of the optimization was to minimize the residual binding due to residual Gd isotopes in the fuel at the end of cycle (EOC). The GA process for creating new BP designs in a coded form called genotypes is generated randomly resulting in a large number of invalid designs. Each new BP design or genotype created by the new GA must be decoded into its corresponding phenotype so that it can be evaluated with a coupled fuel lattice and core depletion calculation. It is essential that most of the invalid designs be eliminated before performing the precise coupled fuel lattice calculation because of the long CPU time that it takes for this calculation. The elimination was accomplished in the new GA by incorporating a beginning-of-cycle (BOC) Kinf filter. The BOC Kinf filter eliminated most of the invalid new genotypes by assigning a high negative penalty to all genotypes that have a BOC Kinf greater than some limit (1.065) for the reference TMI-1 FA. This filter eliminates the need for performing coupled lattice and core depletion calculations for these genotypes. It accelerated the solution process and allowed evaluation of all new genotypes within one day. In this way, the GA minimized the residual binding using an objective function, which maximized the EOC soluble boron (SB) concentration. In essence, the EOC SB or its equivalent EOC keff was maximized.