Materials innovation is an essential piece of driving growth and sustaining industries. However, the process of materials discovery is slow and expensive. Currently employed combinatorial approaches generate a wide range of potential candidates, far exceeding the available capabilities for human evaluation and experimental validation. In this talk, AI-driven molecular inverse-design system for materials discovery coupled with high-throughput experimental validation of predictions will be discussed. The combination of substructure-based feature encoding and molecular graph generation algorithms enables users to develop high-speed, interpretable, and customizable design processes. As an illustration of the technique in action, we apply the proposed workflow to the discovery of novel antimicrobial therapies and acrylic polymers with targeted glass transition temperature (Tg).