Developing new materials has historically been time-consuming. One commonly used approach is material doping, in which given a base material, one can change its properties by substituting some elements with new ones or adding additional elements. Computational material discovery involves searching in a large design space to identify candidates for experimental verification. Recently, it was possible to obtain many electrical and physical properties of materials by density functional theory based first-principle calculation, making it suitable for computational doping-based material discovery. In computational doping, one can substitute some of the atoms in a supercell with dopant atoms. However, the actual positions of the dopant elements within the supercell are not known. In this work, we developed a genetic algorithm for finding the most stable structure of the doped material with the lowest free electronic energy. For each candidate atom configuration, we use the Vienna Ab-Initio Simulation Package to calculate its physicochemical properties, which takes about 15-30 h for a supercell grid of 75 atoms. We did computational doping on SrTiO3 perovskite. Experiments showed that our method can reduce the running time for computational doping by up to 70% compared with exhaustive sampling as commonly used now.