Making holes with the minimum thrust force and torque using a drilling machine is challenging for researchers because of the difficulties in setting input parameters such as the type of drill tool, point of angle, and feeding speed. Therefore, the trial-and-error method to predict optimum input parameters through experiment can be replaced with the Back Propagation Neural Network (BPNN) and metaheuristic method (i.e., genetic algorithm (GA) and Simulated Annealing (SA)) method to reduce costs and time. BPNN can be used to represent the input-output correlation precisely. However, obtaining a model with minimum Mean Squared Error (MSE) requires much data for training, testing, and validation. Since the obtained data from experiments requires expensive costs, combining data from experimental and simulation using ANSYS should be considered to reduce the experimental costs. This study was then conducted to answer the research problem using an EMS 45 tool steel as the workpiece, with the three input parameters: type of drill tools (HSS M2 and HSS M35), the points of angle (118 and 134 degrees) and feeding speed rates (0.07 and 0.1 mm/s). The 32 data from experimental and modeling were used to model the correlation between the input and output parameters of the drilling process using BPNN. The BPNN's network model with minimum MSE is then used as the objective function to determine the input parameters to obtain the smallest value of thrust force and torque using the hybrid method using GA and SA. As the results, the optimum output parameter value in drilling for material EMS-45 (i.e., thrust force = 1615.2 N, and torque = 3.0236 Nm) can be obtained with the following input specifications: (a) type of drill tool = HSS-M2, (b) point of angle = 132.28 o , (c) feeding speed = 0.1 mm/s.