In this study, the aerodynamic behavior of a moderately cambered NACA 4412 airfoil was investigated over a wide range of angles of attack and Reynolds numbers. Computational fluid dynamics (CFD) simulations were performed using ANSYS Fluent with the Spalart–Allmaras turbulence model. The mesh was designed to be uniform with high refinement in critical regions, including the leading edge and the flow separation area. Validation of the CFD results against experimental wind tunnel data showed that the lift and drag coefficients were accurately predicted both before and after stall. Flow field analysis revealed that stall occurrence was caused by early boundary layer separation in the suction region and the formation of a large recirculation zone behind the leading edge, resulting in a sudden drop in lift coefficient and a sharp increase in drag coefficient. Increasing the Reynolds number delayed stall onset and increased the maximum lift. To enable rapid prediction of aerodynamic coefficients, a multilayer neural network was developed with angle of attack and Reynolds number as inputs, and lift and drag coefficients as outputs. The proposed model was able to predict the stall angle with over 90% accuracy. These results demonstrate that the combination of CFD and neural networks provides an efficient and cost-effective approach for predicting airfoil aerodynamic behavior, and can be applied in airfoil optimization, blade performance enhancement, and the development of active stall control systems.