Analysis and design of space structures are generally time-consuming due to the large number of structural members involved. This complexity increases significantly when optimization procedures are required, as numerous analyses must be performed. In this study, artificial neural networks combined with a genetic algorithm are applied to optimize double layer steel grid structures. A total of 180 structural models with three different topologies, spans ranging from 10 to 75 meters, and heights between 1 and 2.5 meters are analyzed and designed for minimum weight. The obtained results are used to train and test the proposed neural network system for predicting structural weight. Subsequently, a neural-network-based genetic algorithm is employed for design optimization. For a site plan with arbitrary dimensions, the neuron–genetic system provides an optimized double layer steel grid configuration, including topology selection, optimal height, column spacing, and horizontal member lengths.