b'FROM QA/QC GRIDLOCK TO MACHINE-LED PRECISIONHow NYCDEP Reclaimed StaffTime and Decision Confidencewith AI-Powered Data CorrectionWhen the System Never Sleeps, The Turning Point: Teaching the Machine Neither Should Your Data Quality to Think Like a HydrologistNew York City runs on water, and a lot of it. Behind every gallonIn partnership with Aquatic Informatics, NYCDEP piloted delivered to over 9 million residents is a system of reservoirs,HydroCorrect, a machine learning-powered QA/QC engine tunnels, diversions, and analytics. But when that system producesbuilt into the Aquarius platform. The tool learns from human inputs, anomaliesspikes, flatlines, or sensor noiseresponse teams canapplying that intelligence across multiple time series, reservoirs, and be overwhelmed sorting signal from noise. recurring error patterns.For the New York City Department of Environmental Protection (NYCDEP), manual review of anomalies wasnt just a hassle, it was a bottleneck. Data confidence was eroding. Skilled staff were spendingWe are working towards a paradigm shift their days chasing false alarms. Decision-making lagged under thefrom humans doing data correction to humans weight of uncertainty. monitoring machines doing data correction.The Challenge: High Stakes, DIRK EDWARDS, ACCOUNT MANAGER, AQUATIC INFORMATICSLow Trust in the Data StreamWith 445 monitoring locations collecting data every five minutes, even minor anomalies multiplied fast. Staff spent up to 20 minutesThe pilot started with just 11 time series, including known problem analyzing each alert, deciding whether to dispatch teams or dismisssites like the Shandaken Tunnel and Rondout Effluent Chamber. the noise. Within two months, NYCDEP fully trusted the system to auto-correct Alerts came up on a dashboard, and staff had to go through theanomalies.data manually to determine whether the anomaly was an issue that required investigating or just poor data.The result? Time lost. Morale drained. And a risk of missing the anomalies that truly mattered.'