P.Eng. / Applied AI / Computer Vision / Technical Workflows

Reeghan Osmond

Applied AI developer and professional engineer building computer vision, 3D imaging, and document workflow systems for industrial, scientific, and regulated technical applications. My work spans Python automation, model development, retrieval workflows, metrology, and decision-support reporting.

Selected projects CV
Reeghan Osmond holding surrogate TRISO fuel particles at Canadian Nuclear Laboratories
Photo: Canadian Nuclear Laboratories

At Canadian Nuclear Laboratories, my work sits between applied AI, imaging hardware, backend-style automation, 3D geometry, metrology, and reporting. I build traceable analysis workflows for segmentation, instance detection, retrieval-supported document work, QA reporting, and technical decision support in regulated research environments.

Selected projects

Applied AI and imaging projects from research, industrial inspection, regulated technical work, and open-source learning.

NuclearFlowAI

In-progress nuclear document workflow demo with cited retrieval, saved findings, summary export, audit logging, usage caps, and deployment-oriented backend planning.

FastAPI Next.js RAG Auditability

Automated 3D XCT analysis for advanced fuel

Applied AI and Python workflows for volumetric segmentation, instance detection, orientation estimation, metrology, mesh export, and QA reporting for nuclear fuel research.

XCT 3D U-Net Python QA reports
XCT scans of TRISO compacts showing particle and kernel distributions
XCT image from CNL

Robotic bin-picking perception pipeline

Industrial perception pipeline using YOLO visible instance segmentation, depth back-projection, object-level geometry features, and a PyTorch ranking model for cluttered scenes.

YOLO RGB-D Point clouds PyTorch
Robotic bin-picking project figure comparing input, ground truth, and prediction
Project report figure

BEV road corridor segmentation

Sensor-fusion learning project converting Argoverse 2 camera, LiDAR, ego pose, and HD map geometry into BEV inputs for segmentation baselines.

Argoverse 2 BEV LiDAR PyTorch
BEV road corridor segmentation prediction examples
Project output figure

Publications and talks

Colourized XCT scan of a surrogate TRISO fuel particle
Photo: Canadian Nuclear Laboratories