Senior Associate SME (LLM Data Annotation and RLHF)
As Senior Associate SME at Innodata, I fine-tuned frontier Large Language Models (LLMs) by creating complex multi-step Biology prompts and solutions. I led the development of evaluation benchmarks and executed detailed RLHF (Reinforcement Learning from Human Feedback) annotations to refine model performance and accuracy in technical domains. My efforts focused on reducing hallucinations and aligning AI outputs with PhD-level biological standards. • Authored Chain-of-Thought (CoT) multi-step reasoning pathways for biology-related tasks. • Provided technical and RLHF annotations to optimize model outputs. • Designed and implemented benchmarks to probe LLM model limitations. • Worked primarily with Qwen, ChatGPT, Gemini, and DeepSeek models.