What

The Python signal-processing and quality-assurance platform behind Kernel Flux: preprocessing, artifact suppression, feature extraction, and performance dashboards over data coming off a 432-channel optically-pumped MEG array.

Why it mattered

The pipeline existed as much for the hardware as for the neuroscience. With a sensor array still under active design, the QA metrics were the feedback signal — per-channel noise, artifact behaviour, and task-locked response quality told the hardware team what to change next. This is where “good data is the bottleneck” stopped being a slogan and became an engineering practice: the software’s first job was to measure the instrument, so the instrument could be improved.

My role

Built the preprocessing, artifact-suppression, and feature-extraction workflows and the performance dashboards, and fed their output back into hardware design decisions.

Outcome

Signal-processing and QA workflows in production use on the Flux system through its development, supporting the human validation studies run on it.