Senior AI Data Strategist
Led large-scale RLHF and Red Teaming campaigns for frontier LLMs to improve model instruction-following and safety. Evaluated and ranked model-generated responses using detailed rubrics, identifying and correcting outputs with hallucinations in code and reasoning tasks. Delivered over 1,000 high-quality annotations weekly while collaborating asynchronously across global teams. • Multi-turn preference ranking and reward modeling for logic, safety, and tone. • Prompt engineering and adversarial prompt testing to surface model misalignments. • Instruction-following evaluation and factuality annotation for LLMs. • Supported SFT dataset creation for domain-specific fine-tuning.