This study offers a hybrid Simulation-Optimization approach combining Monte Carlo simulation (for demand uncertainty) with Bayesian optimization (for policy tuning), an efficient way to optimize inventory decisions compared to brute-force methods like grid search. It compares two key inventory policies, (1) periodic review (p, Q), and (2) continuous review (r, Q) on historical demand data. While both (r, Q) and (p, Q) policies are well-known, there is a lack of empirical studies comparing their performance under real-world demand variability and lead time uncertainty using advanced simulation-optimization techniques. The results show that the (r, Q) policy performs better, increasing expected profit (18.64%) by dynamically adjusting inventory based on daily demand and lead times.