AI Data Expert
I design and develop scalable AI training and evaluation tasks for large language models, focusing on improving reasoning, factual accuracy, and policy compliance. I conduct Supervised-Fine Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) to refine model responses using guided prompts, structured annotation, and dataset validation. I perform multimodal LLM evaluation, safety auditing, pairwise ranking, and red teaming to challenge model logic and ensure robust, safe outputs. • Evaluated LLM outputs for reasoning, correctness, and bias via prompt/response assessments. • Conducted annotation, dataset validation, and alignment checks using standardized guidelines. • Performed SxS and pairwise ranking, safety audits, and hallucination detection on text, audio, and image data. • Used Snorkel AI and Internal/Proprietary Tooling for all RLHF and evaluation processes.