Magic Leap

- Member of the object recognition team focusing on building models for 2D object recognition.- Use pytorch and detectron2 libraries to experiment with and implement state-of-the-art objectdetection algorithms like EfficientDet, DETR, Yolo-v4, and Mask-RCNNs.- Built on R&D work related to improving models with long-tailed distributions and objectsynthetic data simulation techniques.- Managed the team's data gathering and annotation process by working with third party teams toperform data annotation and quality performance checks.