Recycled aggregate concrete–filled steel tubes are composite structural members consisting of steel tubes filled with recycled aggregate concrete produced from crushed construction and demolition waste. The steel tube confinement enhances the mechanical performance of the concrete, increasing load–bearing capacity and ductility under axial compression. While experimental research on these sustainable members is growing, high–fidelity 3D numerical models capable of accurately predicting their behaviour remain limited. This paper presents an advanced framework for simulating the axial response of this type of structural member. The workflow integrates Bayesian optimisation algorithm with a 3D nonlinear finite element model available in commercial software, automated via Python scripting. A dataset of 209 experimental specimens is used to calibrate a constitutive model for confined recycled concrete, essential for reliable column analysis. The proposed model extends the Eurocode 2 stress–strain relationship in compression to capture the specific behaviour of confined recycled aggregate concrete. Validation against experimental results shows good agreement in load–displacement response and ultimate strength prediction. Derived from a wide experimental dataset, the modelling strategy enables rapid and reliable assessment of column performance. Beyond its application to this column type, the proposed framework illustrates the broader potential of integrating simulation, optimisation, and data–driven modelling, offering a transferable methodology for diverse structural problems.