Computationally Intensive STEM AI Trainer (AI solution evaluation and proof-based feedback) at Mindrift
Evaluated AI-generated solutions to advanced STEM and mathematics problems for correctness, logical consistency, attention to detail, and reasoning quality. Verified model outputs against rigorous proof expectations and identified subtle errors or invalid implications. Produced corrected step-by-step solutions and expert feedback suitable for mathematics training and LLM evaluation workflows. • Checked reasoning quality and logical consistency in advance mathematical arguments • Authored and reviewed expert prompts and problem designs emphasizing rigorous reasoning and computation where useful • Repaired flawed solutions with clear mathematical language and LaTeX-style formatting • Used Python-supported computational checks (NumPy/SciPy/pandas) to support verification when appropriate