In a context where magic state preparation dominates the space-time cost of fault-tolerant quantum algorithms, recent advances have substantially widened the factory design space: batched protocols producing several magic states per run, compact protocols leveraging qubit recycling, and "unfolded" distillation performed at the physical rather than the logical level, which exploits the noise bias of cat qubits. Because the repetition codes underlying Alice & Bob's architecture do not natively give access to the full Clifford group, the architecture must also distill Clifford resource states in addition to non-Clifford magic states. Selecting the best factory is a highly multi-parameter optimisation problem, coupling spatial and temporal code distances, single- versus multi-output protocols, the degree of parallelism of the distillation circuit, the magic state injection scheme, the factory layout, and hardware connectivity constraints. The goal of this internship is to navigate this design space and identify the most resource-efficient factories for Alice & Bob's architecture, focusing first on magic state factories and possibly extending the resulting tools to Clifford state factories.