At CVAT.AI, I annotated dense video datasets for action recognition and robotic interaction models, tying event boundaries to the exact frames of first contact and release. I also built a timestamp-checking routine to catch timeline drift before submission.
At iMerit Scholars, I reviewed multimodal and computer vision datasets against client rubrics, audited peer batches, and helped address recurring annotation errors. I’ve also evaluated LLM responses at SkillCat and annotated NLP and computer vision datasets at Avala AI.
