Grid engineers and signal scientists

Grace Chen

CEO, Co-founder

Grace spent years in distribution operations, primarily managing transformer maintenance programs for utilities in the mid-Atlantic and Midwest. The consistent pattern she observed: utilities doing annual oil analysis, receiving results showing elevated dissolved gases or degraded insulation values, then scheduling inspections for six months later. By inspection time, a meaningful fraction of the "watch list" assets had already tripped or were deteriorating faster than the schedule allowed. The data existed. The problem was that conventional monitoring could not see the fault forming in real time. Grace started Magnefy to address that specific gap: not another oil analysis tool, not another SCADA integration, but a system that reads the EM output of the transformer continuously and interprets the signal directly.

Marcus Webb

CTO, Co-founder

Marcus came from relay protection and substation automation, where precise measurement of current and voltage waveforms is standard practice. His observation from that work was that the harmonic content of transformer output, already measured for protection purposes, contained far more diagnostic information than anyone was using for condition monitoring. The protection systems were looking at fundamental frequency quantities and thresholds. The fault signatures he was interested in showed up in the higher harmonics and in the time evolution of the distortion pattern, not in any single-point measurement. He built the first Magnefy prototype sensor to capture that signal, and the initial validation against transformers with known fault histories confirmed that the signatures were reproducible and classifiable.

Nadia Osei

Head of Data Science

Nadia joined after completing graduate work in time-series anomaly detection applied to rotating machinery and industrial equipment. Her background was in vibration analysis for mechanical systems, where the connection between frequency-domain signatures and fault progression is well-established. The intellectual challenge at Magnefy was adapting that framework to electromagnetic output and building a fault library that could generalize across transformer manufacturers, age ranges, and operating load profiles. The current fault signature library covers winding insulation degradation, partial discharge precursors, turn-to-turn fault onset, and core saturation anomalies. Each signature class has both frequency-domain and time-domain features that the classification models use to distinguish fault types and estimate progression rate.