A novel blowing flow control technique, called Coanda jet flap, to improve the performance of the highly loaded axial compressor is numerically and experimentally investigated in this study. A theoretical model is initially built to guide the preliminary design of the jet velocity. The design and effectiveness of the Coanda jet flap are then studied for two different blade designs: a highly-loaded double-arc blade cascade and a highly-loaded axial compressor stator. The optimal aerodynamic shape of the Coanda jet flap is achieved based on the genetic algorithm in conjunction with the artificial neural networks. A comparison of the flow details between the optimization result and the prototype is then conducted to understand the flow mechanisms responsible for the performance enhancement. The numerical and experimental results show that the Coanda jet effectively improves the aerodynamic performance of the highly loaded cascade. The total pressure loss is reduced by a maximum of 22.4% when the normalized mass flow rate of Coanda jet is equal to 1%. Meanwhile, the static pressure rise coefficient is increased by up to 14% with a 1.5% jet mass flow rate. The cantilevered stator vanes with Coanda jet flap is well-designed and embedded in a 1.5-stage highly loaded axial compressor stage. The numerical data shows that the total number of stator blades can be successfully reduced by 35% with the uncompromised overall performance when the Coanda jet flap is applied.