Polymeric materials show advantages of high performance and low cost and have been widely used in modern society. Epoxy resin (EP) as one of the most important polymers, has been widely employed in construction, automotive, and aerospace industries, etc. In Europe, EP market has exceeded 410.60 kilo tons per year. However, the intrinsic flammability of EP makes them potential hazards to human life and property. Due to environmental concerns, incorporating phosphorus-containing flame retardants (P-FRs) into EP has been approved as an efficient approach to improve their fire safety. However, customized target P-FRs design is still hindered due to limited known structures and mechanisms. For example, in EP, two P-FRs with similar chemical structures but different chemical surroundings showed more than 10 times difference in flame-retardant efficiency. Therefore, how to reveal the structure–property relationship of P-FRs and achieve new high-efficiency P-FRs design based on subtle differences in these chemical surroundings (such as group positions, spatial configuration, etc.) pose significant challenges. The overarching goal of the ambitious yet feasible project (FireDesign) isto design novel high-efficiency P-FRs using machine learning (ML) and genetic algorithms (GA). More specific objectives include: O1) to establish the linkage between chemical structure of P-FRs and their flame retardancy efficiency, and to predict target properties of flame-retardant EP by ML; O2) to design optimal P-FRs molecular structures by GA in tandem with ML-based predictive model; O3) to synthesize high-efficiency P-FRs according to the designed molecular structures. FireDesign is a typical multidisciplinary approach requiring complementary expertise from the host (flame-retardant design, fire chemistry, polymer processing) and the researcher (ML, data mining, and GA), contributing to the achievement of “The Materials 2030 roadmap”and“Green and Digital strategies” of EU policies.