The expansion of cities escalates the demand on water utilities amid global water scarcity, making leakage management a critical challenge for water sector sustainability. On this subject, the study introduces an advanced practical approach for estimating nodal leakage through the calibration of decentralized networks using available data. The approach includes conducting a pressure step-test for data collection, employing the DBSCAN algorithm for outlier detection, and employing Bayesian optimization for effective calibration. Hydraulic leakage models, particularly the power and modified orifice equations, are innovatively applied to model and identify existing leaks at each network node through deep comparison. The computational analysis demonstrated efficiency and accuracy in overcoming vast computational complexity and time constraints, with the calibration of a real-life network resulting in a cumulative MSE of 0.892 and an average R² and NSE values near 0.98. Additionally, realistic leakage modeling revealed inaccuracies in the acknowledged connection between the modified orifice equation and the leakage power equation. This study also provides key insights for enhancing water loss management and conducting on-site inspections in an environmentally conscious manner, especially crucial for budget-constrained water utilities and authorities experiencing revenue declines.