A Machine Learning (ML) integrated process was utilized to optimize the properties of Sn–Ag–Cu (SAC) alloys. Bi and In were selected as the key alloying elements to optimize the mechanical properties of SAC387 based alloys. A virtual sample space was constructed for SAC387-xBi-yIn alloys using a machine learning model based on Gradient Boosting Decision Tree (GBDT) algorism. A series of virtual samples were selected and experimentally validated. Sample SAC387-3.5 wt%Bi-2.8 wt%In demonstrated an outstanding mechanical property, yielding a significant improvement in tensile strength (85.8% improvement vs. SAC387, 106% improvement vs. SAC305) with an acceptable ductility (>20%). The synergistic effects of co-additions of Bi and In on the properties of SAC387 based alloys were elucidated. The experimental results show good consistency with the ML predictions, proving that ML can be used as a powerful tool for designing lead-free solder alloys with optimized properties.