Selected work
Work in AI
and software.
Selected projects, from my current work in building automation back to my doctoral research in audio–visual speech recognition.
Honeywell Applied AI 2025 — present
Modelling buildings with graph neural networks
A framework for representing building topology, developed locally and deployed to a governed endpoint on Azure Databricks.
Governed endpoint
Xperi Edge AI 2021 — 2024
Speech recognition on a 100 mW processor
An end-to-end speech recognition engine for the Ergo edge processor, decoding in real time with a fixed one-second delay.
100 mW / 1 s latency
Trinity College Dublin Multimodal research 2016 — 2020
AV Align: learning to see speech
Cross-modal attention for speech recognition, developed during my PhD to learn how sound and lip movement fit together.
Encode → align → decode
My research lives in the open, too.
Publications and open-source work