The design and optimization of mechanically hard soft magnetic materials, which combine high hardness with magnetically soft properties, represent a critical frontier in materials science for advanced technological applications. To address this challenge, a data-driven framework is presented for exploring the vast compositional space of high-entropy alloys (HEAs) and identifying candidates optimized for multifunctionality. The study employs a comprehensive dataset of 1 842 628 density functional theory calculations, comprising 45 886 quaternary and 414 771 quinary equimolar HEAs derived from 42 elements. Using ensemble learning, predictive models are integrated to capture the relationships between composition, crystal structure, mechanical, and magnetic properties. This framework offers a robust pathway for accelerating the discovery of next-generation alloys with high hardness and magnetic softness, highlighting the transformative impact of data-driven strategies in material design.