<p>The aim behind this work is to develop a reliable, fast, and accurate solution for extracting material parameters from Terahertz Time-Domain Spectroscopy measurements needed for characterization and material identification. However, this task is getting increasingly complicated, especially because of challenges such as large data volume, phase unwrapping inaccuracies and external factors like water vapor which can alter the whole performances. To address these issues, we propose a special implementation of genetic algorithm accommodating low signal-to-noise ratios, multiple internal reflections inside the samples and water vapor resonant peaks. Unlike traditional analytical techniques relying on approximations, genetic algorithms provide greater precision. It also outperforms iterative root-finding methods in terms of implementation simplicity and computational speed. The proposed approach is robust against external disturbances and phase unwrapping errors, enabling quasi real-time parameter determination with high accuracy.</p>