Leveraging UMAP for feature selection and Tree Physiology Optimization (TPO) for hyperparameter tuning, we proposed model for efficient water quality control in consumer electronics. We balanced the data with SMOTE from a Kaggle water quality dataset and using UMAP to find important characteristics. The TPO method then was used to maximize the parameters of the deep learning model. With an eye on few false positives, our algorithm was very accurate in spotting clean water. Still, the confusion matrix shows that there are difficulties precisely identifying dangerous water.