In my role as a Data Annotation Specialist for a healthcare technology initiative, I was responsible for meticulously la
In my role as a Data Annotation Specialist for a healthcare technology initiative, I was responsible for meticulously labeling over 5,000 high-resolution retinal fundus images to train a diagnostic Convolutional Neural Network. My daily workflow involved using specialized bounding-box and polygon-segmentation tools to identify and tag microaneurysms, hemorrhages, and hard exudates with pixel-level precision. Beyond raw annotation, I collaborated directly with the machine learning engineering team to refine the labeling guidelines, establishing a consensus protocol for edge cases that reduced inter-annotator disagreement by 15%. This hands-on experience in ground-truth data creation deeply informed my understanding of how data quality directly impacts model convergence and classification accuracy in critical medical AI applications.