Similar to traditional neural architectures, quantum variational circuits face a key limitation in their bespoke, problem-specific designs. Inspired by neural architecture search, quantum architecture search aims to optimize gates and structures within these circuits. This study introduces an innovative quantum search algorithm tailored for NISQ devices which focuses on minimizing the search space, quantum circuit depth and size, while preserving accuracy. The proposal consists of a quantum evolutionary algorithm aided with local search that selects the best performing individuals in a supernet pool. Results exhibit a reduction in the computational footprint, while preserving accuracy.