PFAS pollution is a growing concern worldwide, with no equitable solution in the Thames Basin. We developed a geospatial neural network, predicting PFAS values to within 10% of experimentally validated values. With those predictions, we designed and tested a point-of-use filtration device to be installed on taps. Observing a 93% reduction in PFAS concentration, we reduce PFAS to below health limits of 4 ng/l. Further, we optimised the design, where a 10mm depth of activated carbon allows for 2.5 months of usage, with minimal impact on flow rate and introducing no impurities. Not only do we reduce the devastating impacts of PFAS pollution, but we present a potential solution that is accessible for all.