Abstarct: Drilling, as the only method to access subsurface hydrocarbon resources, is among the most costly operations in upstream oil and gas industries. Consequently, optimizing drilling performance and reducing associated costs have always been of significant concern. The rate of penetration (ROP) is the most critical indicator of drilling performance, directly affecting drilling time and costs, and is influenced by geological, geomechanical, and operational parameters. Among these, controllable drilling parameters such as weight on bit (WOB), rotary speed (RPM), and drilling fluid flow rate play a key role in improving ROP and reducing mechanical specific energy (MSE). Since minimizing non-productive drilling time requires both targeted ROP enhancement and efficient energy consumption, optimization of drilling parameters represents an effective and practical solution. In this study, to optimize controllable drilling parameters for ROP improvement and MSE reduction, geological, petrophysical, and mud logging data from three vertical wells in the Ahvaz oil field were collected and evaluated. First, the geomechanical properties of the reservoir rocks were estimated through a one-dimensional geomechanical model using field-specific correlations, and pore pressure, in-situ stress, and wellbore stability were assessed. Subsequently, drilling and geomechanical data were pre-processed, including outlier removal, normalization, and noise reduction baxsed on wavelet transform, and converted to a unified scale. Next, multicollinearity among variables was examined, and the Non-dominated Sorting Genetic Algorithm II (NSGA-II) was applied to select a subset of features most influential on ROP. baxsed on the selected features, ROP predictive models were developed using machine learning algorithms, including eXtreme Learning Machine (ELM), Multilxayer Perceptron (MLP), and Gradient Boosted Trees (GBT), with GBT selected as the final model due to its superior accuracy and stability. Thereafter, MSE was calculated using the modified Dupriest approach and, together with ROP, incorporated into a multi-objective optimization frxamework. Finally, NSGA-II was employed to optimize the controllable drilling parameters under geomechanical constraints of the formation, determining the optimal WOB, RPM, and drilling fluid flow rate. The results demonstrated that applying the optimized parameters led to increased ROP and reduced MSE in all three wells, achieving reductions in drilling time of 36.09%, 38.76%, and 30.13% for wells 1, 2, and 3, respectively, highlighting the effective role of integrating geomechanical and drilling data in optimizing drilling performance.