Environmental sustainability is a major concern for our societies. This has become relevant also to aviation during the last decades, leading to the development of cleaner means of transportation. The exploitation of propeller-driven aircraft and distributed propulsion is one of the enablers to the transition to sustainable aviation. In particular, propellers can be distributed along the span of the wing to modify the pressure distribution, enhancing the aerodynamic performance of the aircraft. However, the interaction between the wake of the propellers and the wing is complex and may have a negative impact on drag. The optimisation of propellers (shape and position) should then take into account their effect on the aircraft. We propose a multi-fidelity optimisation approach based on the open-source tools DUST and GEMSEO, and apply it to the preliminary design of a distributed electric propulsion system for a regional aircraft within the Clean Aviation JU. Combining models with different fidelities, the propeller-wing system can be simulated with a high accuracy/cost ratio in comparison to high-fidelity CFD approaches. We employ the lifting-line theory to model propellers, and vortex-lattice or panel methods for wings. Wakes are simulated as vortex particles generated at the trailing edge of aerodynamic surfaces. Airfoil characteristics for the propeller blades are computed from cheaper two-dimensional simulations, enabling the use of higher-fidelity methods (TAU/CFD). This approach is integrated into the optimisation library GEMSEO and further enhanced using generative artificial intelligence. Optimisation of the design space for aerodynamic applications requires a large number of simulations. In particular, the exploration of the design space associated with the aerodynamic profiles is accelerated with generative machine learning methods. A generative model is trained with data computed using low-fidelity simulations including wing-propeller installation effects, and it is then used to generate aerodynamic profiles to optimise the wing-propeller system for the reference design conditions.