AI Trainer (Appen, Outlier, Clickworker, Oneforma, Tech Mahindra)
Performed AI training support by generating and reviewing accurately labeled NLP data for model development. Conducted human evaluation of AI-generated responses by checking correctness, language quality, and contextual relevance against given prompts. Assessed outputs for truthfulness, harmfulness, and cultural appropriateness to support safer, more reliable model behavior. • Labeled/training data preparation for NLP algorithms. • Rated AI responses across multiple prompts. • Evaluated truthfulness, harmfulness, and cultural appropriateness. • Identified improvement areas and delivered detailed feedback to developers.