Problem

Reliability across heterogeneous, long-lived sensing systems

How can sensing and learning systems stay reliable when devices, homes, users, and environments keep changing — with missing labels and long-term drift?

Core research questions

RQ list TBD — structure only for now.

What this is not

Not pure building-energy product work; not pure statistical-learning theory as the identity.

Failure modes in the wild

Dropouts, cross-home shift, calibration drift, sparse supervision — placeholders.

Why doctoral work

Systems + learning novelty under real longitudinal constraints — draft later.