Systems such as electronics, cars, computers, and robots are assembled from modular components for specific applications. Photovoltaic reverse osmosis (PVRO) systems, which can be custom-tailored for the water demands and solar properties of particular communities, are an important potential application of modular systems. Clearly, to be financially viable, such systems must be assembled from commercially available components and subsystems (modules). Designing a system from modular components for a specific application is not simple. Even for a relatively small inventory of modular components, the number of possible system configurations that exist is extremely large. For a small community, determining the best system configuration is an overwhelming task due to lack of expertise. This paper presents a modular design architecture that can be implemented on a laptop so nonexperts can configure systems from modular components. The method uses a hierarchy of filters, which can be provided from an expert system, to limit the large design space. Optimization methods and detailed models are then used to configure the location-specific system from the reduced design space. The method is applied here to community-scale PVRO systems and example cases demonstrate the effectiveness of the approach.