Gaze-informed inspection AI
Using eye tracking to capture how expert inspectors visually assess damaged structures, then distilling that attention into saliency models and vision transformers.
See the research →Data Scientist|Applied Research Scientist|Research-to-Industry Lead|PhD in Engineering
currently |
I'm a Data Scientist at GA Telesis, where I build demand-forecasting and Document AI systems — transformer and gradient-boosting pipelines running in production on AWS.
Before industry, I earned my PhD at Penn State, teaching machines to see the way expert inspectors do — using eye tracking, drones, and deep learning to assess buildings and bridges. That research thread is still very much alive.
If you're working on hard problems where models have to survive contact with the physical world, let's talk.
models, coastlines, and cups of tea
Using eye tracking to capture how expert inspectors visually assess damaged structures, then distilling that attention into saliency models and vision transformers.
See the research →Drone-based visual inspection with geo-tagged capture and deep CNNs — built on data from 50+ inspection flights, turning slow manual surveys into instant, repeatable assessment.
See the project →Production forecasting for intermittent, hard-to-predict demand — transformer and gradient-boosting ensembles on AWS, delivering probabilistic forecasts to decision-makers daily.
See the project →Working on forecasting, computer vision, or infrastructure AI? Hiring for applied science? Or just want to talk shop about drones and models — I'd love to hear from you.
Get in touch