Generalist
Image
Bounding BoxScope & Tasks: Focused on multimodal AI training, specifically the evaluation and refinement of AI-generated video and imagery. My tasks involved performing side-by-side comparisons of model outputs, bounding box/masking for object identification, and providing detailed analytical justifications for ranking responses based on visual fidelity and prompt adherence. Project Size: Contributed to a high-volume, long-term initiative involving thousands of daily tasks, requiring high adaptability to weekly-changing rubrics and technical guidelines. Quality Measures: Consistently maintained high accuracy scores (targeting a 4.0/5.0+ average) by adhering to strict Manual Audit standards. I ensured all justifications were free of AI-generated patterns and focused on identifying subtle visual hallucinations or logic errors in the model’s performance.