Multimodal Data Labeling & Model Evaluation
Internal Proprietary ToolingComputer Code ProgrammingQuestion AnsweringText Generation
Worked on Outlier’s Aether programs performing multimodal data labeling and model evaluation tasks for AI training and quality improvement. Responsibilities included rating model outputs for instruction-following and truthfulness, writing corrected “ideal” responses, and producing transform-based visual reasoning solutions (crop/zoom, dewarp/perspective correction, contrast enhancement, thresholding, morphology, ROI masking, connected-components counting) to solve challenging image questions. Followed strict rubrics and pass/fail requirement checks, provided detailed, evidence-based explanations, and applied quality controls to reduce partial-compliance errors and improve annotation consistency.
Worked on Outlier’s Aether programs performing multimodal data labeling and model evaluation tasks for AI training and quality improvement. Responsibilities included rating model outputs for instruction-following and truthfulness, writing corrected “ideal” responses, and producing transform-based visual reasoning solutions (crop/zoom, dewarp/perspective correction, contrast enhancement, thresholding, morphology, ROI masking, connected-components counting) to solve challenging image questions. Followed strict rubrics and pass/fail requirement checks, provided detailed, evidence-based explanations, and applied quality controls to reduce partial-compliance errors and improve annotation consistency.
2025