Total: 1
High-quality single-photon sources, realized through saturable emitters, photon blockade, or heralded pair generation, are indispensable building blocks for photonic quantum platforms. Although these mechanisms suppress multiphoton emission through distinct principles typically captured by analytical models, their practical implementation is constrained by conflicting requirements for purity, brightness, and indistinguishability, which must be balanced within high-dimensional design landscapes. Here, we propose a computational framework for optimizing competing metrics of single-photon sources. Building on a Liouville-space adjoint formulation that efficiently evaluates multiple objectives in Markovian open quantum systems, we develop a Jacobian-based update, which ensures first-order monotonic reduction of multi-objective costs. By incorporating simulated annealing to escape gradient-vanishing plateaus, our framework achieves a design success rate of nearly 60 % for photon blockade with g2(0) smaller than 0.1 and theoretically bounded brightness across a broad parameter space, without any analytical guidance. This framework provides a general recipe for multi-objective design of open quantum systems.