At Linamar Corporation, I built an MVP multi-agent AI maintenance assistant for robot technicians that retrieves procedures, error-code context, and prior fixes, then generates maintenance reports. I also developed an evaluation harness using Langfuse, LLM-as-Judge scoring, and Human-in-the-Loop review.
At Robarts Research Institute, I'm extending SPIMquant with a PyTorch 3D U-Net segmentation model for light-sheet microscopy brain volumes. I'm also contributing statistics and distributed-processing modules to produce atlas-mapped results for neuro research.
With Western Engineering Biomedical Club, I led a software team building MustangMotion, a smart ACL rehabilitation sleeve. I built its FastAPI and Python backend to process IMU and EMG streams, and designed an ML pipeline for a Recovery Index.
With Western AI, I led development of NeuralApnea Triage, an ECG-based AI system for sleep apnea screening. Its TensorFlow CNN-Transformer achieved 89.21% accuracy and 95.60% AUROC on the held-out test cohort, and I built a Grad-CAM explainability agent for the clinician-facing application.
