A novel reverse osmosis or nanofiltration system (RO/NF) capable of detecting and responding to onset of fouling within the system utilizing uniquely configured membrane permeate flow path within the system which generates a time-sensitive data. Membrane performance data in real-time operating conditions is then utilized for rapid detection of membrane fouling, fouling rate, and cause of fouling, followed by controller-based system generated actions to stop, and recover from fouling or slow-down fouling, and, if required, to predict, plan, and schedule operator intervention steps to recover optimum system operating conditions. The end-result is a novel energy-efficient and fouling-managed advanced (machine learning) reverse osmosis system for brackish water desalination.