The development of environmentally sustainable coatings requires advanced multiscale and multiphysics modeling approaches to bridge the gap between molecular interactions and macroscopic material performance. The MOZART project, funded by the EU Horizon 2021 program, focuses on replacing hazardous Hard Chromium coatings with Safe and Sustainable by Design (SSbD) alternatives. While nickel matrix nano-composite coatings offer a promising solution, their conventional synthesis relies on boric acid, which contradicts sustainability principles. To address this, we have done machine learning-driven modeling to identify sustainable alternatives. A Conditional Variational Autoencoder (CVAE) is used to analyze and generate molecular representations, while Genetic Algorithms explore the CVAE’s latent space to optimize candidate molecules. The methodology incorporates multiscale simulation techniques, allowing for the assessment of molecular properties, reaction mechanisms, and macroscopic coating performance within a unified framework. By combining machine learning, computational chemistry, and material modeling, this research contributes to multiscale multiphysics simulation strategies, enabling the rational design of next-generation sustainable coatings with improved performance and reduced environmental impact.