An optimization study was conducted to find the optimum operational characteristics of a synthetic jet actuator (SJA) to postpone the static stall separation over an SD7003 airfoil at Reynolds number of 60,000. A genetic algorithm (GA) coupled with an artificial neural network (ANN) was employed. Aerodynamic performance (L/D) was chosen as the objective function. Both tangent to the boundary layer (TBL) and the cross to the boundary layer (CBL) configurations of SJA were used and their effectiveness in separation control were compared. The following design variables of the SJA were allowed to change within a predetermined range: location, the opening length, the injection velocity amplitude, the injection angle, and the nondimensional frequency. It was found that for location, opening length, and velocity amplitude ratio, a narrow range near the peak optimum values achieved the best performance. However, for the nondimensional frequency and jet injection angle, the optimum values providing highest performance were in a wider range of values. Activation of SJ actuator improved the aerodynamic performance of the airfoil significantly. However, TBL configuration of SJA produced superior improvement in aerodynamic performance. The optimum aerodynamic performance achieved by TBL-SJA was 34.4, in comparison to 25.3 for CBL-SJA and 5 for the uncontrolled stalled airfoil at 13 deg angle of attack.