receiving a first runtime record, wherein the first runtime record includes radio frequency (RF) signal data collected in a physical space, wherein the RF signal data comprises characteristics of interactions by one or more RF signals with a physical element in the physical space;

processing the first runtime record using a machine learning (ML) model, comprising:

processing the first runtime record using a shared portion of the ML model to generate a set of features, wherein the shared portion is shared by a plurality of basis components of the ML model; and

processing the set of features using the plurality of basis components of the ML model to generate a plurality of inferences;

aggregating the plurality of inferences to generate a prediction comprising a plurality of coordinates; and

outputting the prediction, wherein the plurality of coordinates indicates a location of the physical element in a physical space and wherein the physical element is an RF passive physical element.