AI Trainer / Data Annotator | Freelance
Served as an AI Trainer / Data Annotator for LLM and multimodal model training, evaluation, and alignment. Assessed model outputs for factual accuracy, tone, and logical flow, then provided detailed rationales to support ranking and behavioral fine-tuning. Built prompt sets to test reasoning quality in architecture, urban planning, and creative writing use cases. • Evaluated thousands of AI-generated responses for accuracy, coherence, and safety • Performed bias detection and flagged hallucinations and safety violations • Authored ranking rationales to fine-tune model alignment • Designed complex multi-step prompts to optimize response relevance by ~20%