Research · publications
Research direction: AI systems that survive being used.
Two IEEE conference papers on AI-driven water resource management. The throughline is the same one that runs through the rest of my work: a model is not the deliverable; an operator-usable, accountable loop is.
Conference publications.
-
01
IEEE Conference · 2024
AI-driven monitoring and decision support for urban water resource systems
Proposes an end-to-end pipeline that fuses sensor telemetry from distributed water-network nodes with a learned demand / leakage model and exposes the result as an operator-facing decision support layer. The contribution is the integration story — collection, modeling, and surfacing — evaluated on a real network rather than in simulation only.
-
02
IEEE Conference · 2024
Operationally accountable AI: keeping the loop honest under real-world use
Builds on the first paper to argue that an isolated model is not the unit of evaluation for water-system AI; the unit is the loop of data → model → decision → operator action → ground truth. Presents the instrumentation we used to keep that loop accountable when the system is actually used.
Why water-resource systems specifically.
Water networks are a useful proving ground for AI-systems work: distributed sensing with real noise, decisions with real consequences, operators who hold tacit knowledge the model needs to respect. It's a domain where you can't ship a model that only looks good on a held-out test set — the field will surface every gap.
The same architectural instincts that show up here — schema-first intake, explicit state, an operator-facing layer that isn't an afterthought — are the ones I bring to product systems too.
Long arc: silicon-based III-V, heterogeneous integration.
Beyond the published work, I keep a reading track on semiconductor-aware engineering — silicon-based III-V devices, heterogeneous integration, and the lower-layer constraints that start to shape software decisions when you push close enough to the metal. This isn't a publication track yet; it's where the longer-term direction is heading.