Mathematician - AI Reasoning & Agent Systems at Mercor

Mathematician | $53–74/hr | Remote Worldwide

Mercor is hiring mathematicians on behalf of a leading AI Lab developing advanced agent AI systems capable of complex reasoning, planning, and autonomous problem-solving. This is a unique opportunity to work directly with world-class AI researchers and engineers, contributing the mathematical foundations that power next-generation intelligent agents.

Responsibilities

• Develop rigorous mathematical tasks, proofs, and structured reasoning datasets used to train and evaluate agentic AI models
• Analyze model reasoning steps to identify logical weaknesses, conceptual gaps, and opportunities for improved mathematical understanding
• Collaborate closely with research teams to design multi-step problem-solving pipelines, evaluation benchmarks, and reasoning curricula
• Create high-quality annotations, explanations, and walkthroughs that teach models how to solve mathematical problems step-by-step
• Support model training efforts by constructing mathematical tests for generalization, abstraction, deductive reasoning, and symbolic manipulation
• Work with engineers to integrate mathematical constraints, verification tools, and logic-based checks into agent workflows
• Participate in synchronous collaboration sessions (4-hour windows, 2–3 times per week) to refine reasoning approaches, review agent outputs, and iterate on dataset improvements

Requirements

• Strong background in Mathematics, Applied Mathematics, Theoretical Computer Science, or a closely related quantitative discipline
• Deep understanding of core mathematical fields: algebra, calculus, probability, combinatorics, optimization, logic, or number theory
• Demonstrated problem-solving ability through research, olympiads/competitions, teaching, or advanced coursework
• Ability to break complex problems into clear, structured steps suitable for model learning and reasoning supervision
• Familiarity with AI reasoning models, LLMs, or agentic AI concepts is a strong plus
• Comfort working with Python for inspecting model outputs, running experiments, or generating mathematical datasets
• Excellent collaboration and communication skills
• Ability to commit 20–30 hours per week, including required synchronous collaboration sessions

Why Join

• Work directly with a world-class AI research lab building the reasoning capabilities behind tomorrow's autonomous agent ecosystems
• Influence how advanced AI systems learn mathematics, apply logic, and develop reliable multi-step reasoning abilities
• Enjoy schedule flexibility — select your own 4-hour collaboration windows and manage your 20–30 hour work week
• Be engaged as an hourly contractor through Mercor, giving you autonomy while contributing to mission-critical agent reasoning projects
• Collaborate with top mathematicians, researchers, and AI developers working at the intersection of formal reasoning and advanced intelligence
• Join a global network of technical experts shaping how the next generation of AI agents think, solve problems, and evolve at scale