The increasing level of congestion and infrastructure costs in cities have created a need for more intelligent and flexible transport systems. Urban Air Mobility (UAM) introduces the third dimension to intra-urban transport at a minimal infrastructure cost, bypassing congestion and providing reliable travel times to users through the provision of air passenger transportation. The performance of UAM systems is highly dependent on vertiport locations, vehicle sizing and infrastructure specifications, which themselves are intrinsically linked. This study takes a holistic approach to UAM network optimisation by considering the inter-relatedness of these decisions in a three-stage algorithm. In the first stage, a linear vertiport placement model with vehicle sizing constraints is developed to determine the optimal vertiport configuration while considering eVTOL performance. The vertiport configuration serves to determine the operational requirements of the aircraft and are incorporated in the vehicle sizing models. The resulting network configuration and vehicle sizing constraints are used as inputs in the infrastructure model, which utilises open network theory to determine the service rate requirements and the allowed loiter and waiting times based on the number of take-off, landing, and charging pads at each vertiport. These stages are executed sequentially through a feedback mechanism, which balances the infrastructure, operational costs as well as passenger waiting times. The algorithm optimises the profitability of the UAM network ensuring all operational constraints are satisfied. A case study based on the hypothetical implementation of UAM in the city of London is presented using Rolling Origin and Destination Survey data to estimate demand patterns. Our results suggest that ignoring vehicle sizing and infrastructure modelling in the network optimisation stage results in infeasible UAM networks. Furthermore, reducing waiting times and loiter times are critical to reduce operational costs of the UAM network, with all optimal configurations yielding waiting times below 5% of the total flight time. The proposed method can be used to plan future UAM developments whilst ensuring operational viability.