How a Mathematics Professor is Shaping AI Reasoning

From Lecture Hall to AI Lab

When James Whitfield first heard about IXO, he was skeptical. As an Assistant Professor of Applied Mathematics at a major research university, his days were already packed with teaching, publishing, and advising graduate students. The idea of taking on additional work seemed impractical.

"I thought it would be another low-quality gig platform," James recalls. "But when I saw the caliber of projects — evaluating large language models on complex mathematical proofs, refining reasoning datasets for frontier models — I realized this was different."

The Work That Matters

James's expertise in abstract algebra and topology makes him uniquely qualified for a specific niche in AI training: mathematical reasoning evaluation. His work at IXO involves:

• Evaluating LLM outputs on graduate-level mathematics problems, identifying where models succeed and where they hallucinate
• Creating high-quality training data for mathematical reasoning, including step-by-step proof annotations
• Contributing to RLHF pipelines that help models distinguish between valid and invalid mathematical arguments

"The quality of projects at IXO is remarkable," James says. "I've evaluated large language models, refined mathematical reasoning datasets, and contributed to RLHF pipelines — all while maintaining my full-time research position."

Impact on AI Development

James's contributions have directly influenced how leading AI labs approach mathematical reasoning in their models. His annotations on over 2,000 mathematical proofs have been used to fine-tune models that now demonstrate significantly improved performance on graduate-level mathematics benchmarks.

"Every annotation I provide isn't just a data point — it's a piece of mathematical knowledge that helps AI systems reason more rigorously. That's genuinely exciting for a mathematician."

Balancing Academia and AI Training

The flexibility of IXO's platform has been crucial for James. He typically works on IXO projects during evenings and weekends, fitting tasks around his academic schedule.

"The compensation reflects the specialized knowledge required," he notes. "I earn more per hour on IXO than from most consulting opportunities, and the work is intellectually stimulating in a way that complements my research."

Advice for Fellow Academics

James encourages other academics to consider expert AI training work:

1. Your expertise is valuable — AI labs need domain specialists, not just ML engineers
2. The flexibility is real — no minimum hours, no fixed schedules
3. It enhances your research — understanding how AI models handle your domain gives you unique insights

"If you're an expert in any field, there's almost certainly a project at IXO that needs your knowledge. The AI revolution needs domain experts as much as it needs engineers."