<p>Accurately estimating Pile Bearing Capacity (PBC) is crucial for safe, economical pile foundation construction, ensuring optimal material use and enhanced structural integrity. The study focuses on developing a framework using the Multi-Layer Perceptron (MLP) model to predict PBC. MLP is selected for its effectiveness in modeling complex, non-linear data links. To enhance the MLP's performance, two optimization algorithms, African Vulture Optimization (AVO) and Reptile Search Algorithm (RSA), are integrated. These algorithms are selected for their ability to balance exploration and exploitation, helping the MLP avoid local minima and achieve global optima in complex optimization tasks. The proposed AI models achieved remarkable accuracy, with the MLPS2 (MLP + RSA in the second layer of MLP) and MLAV2 (MLP + AVO in the second layer of MLP) models attaining substantial Coefficient of Determination (R²) values of 0.995 and 0.988, alternatively. The MLPS2 model also obtained Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) values of 126.70 KN and 102.80 KN, respectively. The MLAV2 model was demonstrated as the second-best model among the proposed frameworks, with an RMSE value of 203.04 KN. The developed AI predictive frameworks based on MLP and hybrid optimization techniques effectively enhance PBC estimations, paving the way for improved pile design and optimization.</p>