At Mindy Support, I annotated video frame by frame using bounding boxes and semantic segmentation polygons to label spatial environments for robotics and computer vision models.
I also wrote structured captions and timestamp metadata for visual actions and object interactions. I reviewed annotations for pixel-level discrepancies, documented edge cases, and maintained a reported accuracy score above 98% against client labeling guidelines.
At Remotask and Outlier AI, I evaluated AI-generated video summaries, text responses, and language scenarios for factual accuracy and alignment with prompt constraints. I documented failure modes and provided structured feedback to support prompt engineering.
At National Cereals and Produce Board, I handled routine technical support and assisted teams with technology implementations. I also built a personal portfolio project demonstrating workflow optimization, presentation structures, and data handling processes.
