Data Annotator
Project 1: Multimodal Fashion Image Transformation (Aether Project Ecosystem) Project Description: Contributed to generative AI model training within Scale AI's Aether project, identifying complex visual attributes in fashion datasets to direct automated style and pattern transformations. Data Labelling Tasks: Utilized the proprietary Multimango web interface to process and select multi-item fashion outfits from raw image datasets. Mapped and segmented visual boundaries of clothing items. Applied text-to-image prompts and attribute tags to execute precise pattern transformations (e.g., solid to plaid, floral to geometric). Evaluated the realism and accuracy of the resulting AI-generated transformations. Project Size: Large-scale dataset pipeline involving thousands of image assets per batch cycle across a global contributor network. Project 2: Multimodal Animation Video Interpretation (Aether Project Ecosystem) Project Description: Enhanced video-to-text generative models by providing high-quality, descriptive training data for animated video sequences within the global Aether project initiative. Data Labelling Tasks:Reviewed and analyzed short-form animation clips using dedicated web-based tools. Generated dense, granular textual descriptions of character actions, emotions, and environmental changes. Timestamped specific visual events to align video frames with textual metadata. Classified animation styles, frame-rate consistency, and visual continuity errors. Project Size: High-volume, continuous stream of multimedia content requiring rapid, precise turnaround times.