Wastewater testing has emerged as an effective, widely used tool for population-level SARS-CoV-2 surveillance. Such efforts have primarily been implemented at the wastewater treatment plants (WWTPs), providing data for large resident populations but hindering the ability to implement targeted interventions or follow-ups. Conversely, building-level wastewater data exhibits increased variability due to rapid daily population dynamics but allows for targeted follow-up interventions or mitigation efforts. Here, we implemented a three-site wastewater sampling strategy on our university-affiliated medical campus from May 2021 to March 2024, comprised of two distinct hospital quadrants and a primarily research laboratory and classroom building. We first addressed several limitations in implementing hospital-level wastewater surveillance by optimizing sampling frequency and laboratory techniques. We subsequently improved our ability to model SARS-CoV-2 case counts using wastewater data by performing sensitivity analyses on viral shedding assumptions and testing the utility of internal normalization factors for population size. Our unique infrastructure allowed us to detect intra-hospital dynamics of SARS-CoV-2 prevalence and diversity and confirmed that direct sequencing of wastewater was able to capture corresponding clinical viral diversity. In contrast, research building wastewater sampling showed that for most non-residential settings, despite low overall viral loads, a threshold approach can still be used to identify peaks in cases or transmission amongst the general population. Our study expands on current wastewater surveillance practices by examining the utility of and best practices for upstream and particularly hospital settings, enabling the use of non-municipal, medium-scale wastewater testing to inform efforts for reducing the burden of COVID-19.