ikeGPS
Constructing advanced ML pipelines to assist in automating critical infrastructure analysis. Pioneering the use of deep learning and computer vision to revolutionize utility pole inspections at scale.
- Engineered a high-precision one-shot object segmentation system achieving ~0.88 mAP for complex utility asset detection
- Developed an end-to-end neural network training suite for automated prediction artifact collection and model lifecycle management
- Created a custom web-based image dataset masking platform with hierarchical object support and role-based access control to accelerate dataset annotation and model training workflows