AI Generalist - Multimodal Video Data Annotation & Evaluation
As an AI Generalist at Aether Project, I performed multimodal annotation on video data with a focus on Named Entity Recognition (NER) and relationship annotation utilizing structured schemas. I conducted qualitative and stylistic evaluations for vision models by comparing outputs and referencing artistic style criteria. My work involved object-level editing on video and image data to ensure precision and quality for downstream model training tasks. • Tagged entities and relationships for NER and contextual annotation across dynamic video content. • Conducted vision model evaluations to assess quality, relevance, and output consistency. • Completed video artistic style reference tasks and object-level editing (e.g., furniture removal). • Ensured guideline adherence, ethical data handling, and provided actionable feedback for model improvement.