High-fidelity simulations are an essential tool for understanding complex aerodynamic phenomena. However, their accuracy comes at a cost in terms of computational resources and execution times. These limitations are particularly restrictive in contexts where rapid evaluation of multiple designs is required to accelerate the design and optimisation process. The advent of data-driven methods and machine learning has opened up new perspectives for attempting to overcome this limitation. As part of this broader shift, reduced-order models (ROMs) have gained particular importance as a means of drastically reducing computational costs while enabling near real-time analysis and simulation of physical phenomena. This allows engineers to perform rapid design iterations, conduct sensitivity analyses, and optimise product configurations more efficiently.