Audio Trainer
-Performed daily preference ranking of hundreds of multimodal AI outputs (images, audio, and video) against prompts to evaluate model performance and alignment. -Applied systematic comparative judgment techniques to assess relevance, quality, and coherence of generative model outputs, ensuring consistent evaluation across diverse modalities. -Maintained high accuracy and throughput in large-scale annotation workflows, contributing to reliable training data for model fine-tuning and preference optimization. -Collaborated with cross-functional AI teams to identify quality trends and provide actionable feedback for improving multimodal generative capabilities. -Utilized specialized labeling platforms and guidelines to ensure precise annotation, supporting preference-based reinforcement learning and human feedback loops.
