Crafted complex prompts across 3 expert domains (Physics, Mathematics, Mechanical Engineering) resulted in 2000+ training hours.
Ranked AI responses using 5+ metrics (accuracy, quality, helpfulness, harmfulness, and compliance with user and system requirements).
Assessed large language models performance for up to 4 responses and provided detailed feedback.
