This research proposes the systematization of evolutionary design processes through a methodology applied to the design of the decentralized headquarters of the Center for Research, Innovation and Development of Aquaculture and Oceanographic Sciences of the Institute of the Sea of Peru (IMARPE), in which establishes a definition of the architectural problem, which covers requirements and restrictions, to later be translated into geometric modification instructions interpreted by an evolutionary algorithm based on machine learning, thus obtaining as a result the volumetric composition of the preliminary project, product of the evolution of 5000 different proposals. , obtained generatively, where the resulting product is the best possible based on the restrictions and requirements set forth. This work is a starting point for generative design methodologies, investigates other ways of approaching design in the teaching of local architecture, and explores the possibilities offered by the evaluation of multiple solutions to a particular problem.