Multimodal AI Data Annotation & Quality Evaluation
Worked on multimodal AI training tasks involving text and audio data annotation, evaluation, and validation. Responsibilities included reviewing sequential audio recordings (system, user, and model responses), ensuring full comprehension before proceeding with annotation tasks, and assessing response quality based on accuracy, relevance, coherence, and instruction adherence. Performed response ranking between multiple model outputs, identified errors and inconsistencies, and flagged edge cases to improve model performance. Maintained strict adherence to detailed annotation guidelines while ensuring consistency across tasks. Contributed to improving training data quality by providing structured feedback and accurate labeling across multiple datasets under tight deadlines.