Senior Domain Expert | Mercor (LLM output evaluation and RLHF-related auditing)
Evaluated and optimized frontier LLM outputs by engineering prompts, identifying edge cases, and auditing responses for strict factual accuracy, logic, and formatting. Produced structured analytical justifications to rank model performance and inform RLHF training pipeline decisions. Performed multi-domain workflow audits to maintain data integrity and quality-assurance metrics. • Engineered complex prompts for improved LLM output quality • Audited responses against factual, logical, and formatting constraints • Authored ranking rationales used to guide RLHF training • Maintained data integrity across multi-domain generalist workflows