<p>Polyester, the most widely used synthetic fibre, has been challenging to dye sustainably with its hydrophobic surface and low reactivity, necessitating energy- and resource-intensive high-temperature high-pressure (HTHP) dyeing processes. Enzymatic surface modification with esterase was investigated as a green method to enhance the dyeability of polyester with reactive dyes. Development of -OH and -COOH groups was successfully confirmed by FTIR. A machine learning model, Extreme Gradient Boosting (XGBoost), was trained to predict the concentrations of -OH and -COOH groups (in mmol/g) and the color strength (K/S value) for modified polyester. A data set consists of 351 experimental data points, each experiment repeated 3 times, comprising process parameters (esterase (%), time, temperature, and pH), was used to train and evaluate the model. Strong predictive capability was demonstrated by the model, with R² values exceeding 0.9 for -OH, -COOH, and K/S. Interpretability by SHAP analysis identified enzyme concentration as the most influential factor, followed by pH, time, and temperature. Residual analysis confirmed normally distributed errors and small deviations of approximately 2 mmol/g for more than 90% predictions, affirming the robustness of the model. Improvement in wicking height, absorbency, and reduced contact angle demonstrated enhancement in hydrophilicity, without affecting the tensile strength. Thermal analysis using DSC revealed melting as well as crystallisation behaviour modifications, complementing modifications at the surface level. Overall, this integrated (enzymatic processing and machine learning) approach for a cleaner production strategy reduces energy usage, minimizes environmental discharge, and presents a viable path toward more sustainable, eco-efficient textile dyeing.</p>