In this research paper, we have employed Multi-Layer Perceptron (MLP) and Kolmogorov–Arnold Network (KAN)-based neural network models to solve the inverse kinematics problem of a 3-DoF RRR articulated manipulator. The dataset was generated using a sinusoidal function to produce joint angles, which were then used to calculate the end-effector position through forward kinematics. Both MLP and KAN-based models were trained and tested on the generated dataset. Experimental results demonstrate that the MLP model achieved a Root Mean Square Error (RMSE) of 0.229 and an R-squared (R2) score of 0.958. In comparison, the KAN model exhibited superior performance with an RMSE of 0.035 and an R2 score of 0.999. These findings indicate that the KAN model outperformed the traditional MLP approach in solving the inverse kinematics problem for the given manipulator configuration.