AI Response Evaluation & NLP Text Annotation for LLM Training
Worked as a Remote AI Data Annotator and Evaluator contributing to the training and alignment of Large Language Models (LLMs). The project involved evaluating, labeling, and improving AI-generated text to enhance model accuracy, safety, and contextual understanding. Scope of Work: Evaluated AI-generated responses for factual accuracy, logical consistency, coherence, neutrality, and completeness. Applied structured evaluation rubrics to rate responses across multiple quality dimensions. Identified hallucinations, misleading content, unsafe outputs, and policy violations. Performed Reinforcement Learning from Human Feedback (RLHF) tasks by ranking multiple model outputs based on quality and alignment. Conducted sentiment analysis, intent classification, and topic categorization. Performed Named Entity Recognition (NER) tagging on legal, general knowledge, and conversational datasets. Assisted in supervised fine-tuning (SFT) by writing high-quality prompt-response pairs.