Abstract : This thesis aims to shed light on the merits of integrating machine-learning (ML) with wastewater treatment processes. An integration, if utilized, will set the inception of next-generation wastewater treatment plants (WWTPs) by dramatically changing the way they operate. The research gap addressed in the thesis is that it investigates the various cases of testing the performance of the ML prediction models against different WWTPs than the one used to train and test. In addition, it discusses the biological factors of wastewater that lead to complications in ML modeling and prediction. Moreover, it incorporates the biological treatment environment constraints with ML models. All these points were never discussed or considered in ML literature. The thesis exploits unprecedented insight into the Kingdom of Bahrain's wastewater characteristics, pattern detection, prediction, and optimization by integrating several ML algorithms on four WWTPs around the Kingdom: Muharraq, Madinat Salman, Askar, and Al Dur WWTPs. A total of 299 tuned ML models were developed for anomaly detection, prediction of influent biochemical oxygen demand (BOD5), optimizing dissolved oxygen (DO) requirement for biological treatment, and prediction of treated sewage effluent (TSE) quality. The anomaly detection results recommend a univariate rather than a multivariate approach for improved performance. On the other hand, the influent BOD5 ML prediction model accelerated the testing results duration from conventional testing procedures by 40 times, reducing it from five days to only three hours, achieving excellent prediction results of 1.00 coefficient of determination (R) and 0.08 mean absolute error (MAE) against Askar WWTP testing dataset. At the same time, it scored 0.94 R2 and 6.56 MAE against Madinat Salman WWTP testing dataset. Conversely, the DO requirement optimization model achieved 1.00 R2 and 0.01 MAE, leading to a 23% reduction in energy and operational costs of air-blowers. Moreover, TSE BODs and COD were predicted with 0.77 R2, 0.07 MAE and 0.68 R2, 0.04 MAE, respectively. All proposed ML model's performance was verified using K-fold cross-validation and interpreted using Shapley Additive Explanation. The verification results proved the model's robustness. These models can be deployed in an integrated framework of remote sensing and ML to improve wastewater treatment processes, reduce energy consumption and operational costs, optimize plant performance, and enhance environmental preservation by suppressing TSE quality violation risks. This is the first research of its kind to be applied to Bahrain's WWTPs.