Selected Work
4 projects in Computer Vision
P&ID Digitization
Scottish Water
Multi-stage computer vision pipeline to extract structured data from thousands of piping and instrumentation diagrams, enabling programmatic query of a legacy drawing library.
AI Culvert Detection
USGS 3D Hydrography Program
Deep learning model to automatically detect and classify culverts from aerial imagery and LiDAR derivatives, supporting national hydrography data modernization.
Post-Disaster Imagery Ontology
NVIDIA GTC / Internal R&D
Knowledge graph and ontology for post-disaster aerial imagery, enabling structured reasoning over damage classifications presented at NVIDIA GTC 2024.
Refinery Oil Seep Detection
Marathon Petroleum — Lovell Refinery
Multispectral and thermal drone imagery analysis to detect subsurface hydrocarbon seeps around a petroleum refinery, reducing manual survey time significantly.