Computer Vision Project Manager
Current- Collecting and managing large quantities of contextual data and labeled roadside-pedestrian footage alongside >300k image COCO dataset using efficient data pipeline architectures
- Deploying the YOLOv8-Pose estimation model with Docker/Kubernetes to semi-autonomously annotate footage on CVAT
- Training heuristic rules-based & Keras deep learning models to classify pedestrians hailing taxis in real-time (5-15 fps)
- Leveraging an OpenAPI-powered NLP labeling-model to communicate hailer positions with 93% mean Average Precision