At xAI, I scored AI-generated responses for accuracy, clarity, reasoning strength, and safety alignment. I also created prompts and responses for SFT datasets and contributed to evaluations involving text, audio, and images.
At Outlier, I reviewed datasets and annotation tasks for accuracy, consistency, and guideline compliance. I maintained a quality score above 90% and flagged labeling inconsistencies and logical errors.
At Erud AI, I annotated and categorized machine-learning datasets, including intent classification, semantic tagging, sentiment evaluation, and dialogue analysis. I also identified structural, contextual, and conceptual errors in training data.
At RWS Group, I reviewed machine-generated outputs for translation quality, classification accuracy, fluency, and alignment with linguistic, cultural, and safety guidelines. Most recently, at Turing, I performed side-by-side evaluations on GHD, fact-checked outputs, and documented the reasoning behind my evaluations.
