External Research Collaborator, Formal Mathematics – AI

This is a fully remote position, open to applicants in France.

📋 Description

• Design, scale, and maintain multi-agent AI pipelines that convert intricate mathematical texts into validated Lean 4 code.

• Assess pipeline performance and troubleshoot verification and compilation issues.

• Develop engineering solutions to enhance the reliability of autoformalization.

• Investigate and implement automatic tactic generation and domain-specific proof-search methodologies.

• Enrich and organize the client’s dataset by formalizing absent mathematical results, theorems, and proofs.

• Perform peer reviews of AI-generated Lean statements and proofs.

• Safeguard mathematical fidelity, proof integrity, logical soundness, and code quality.

• Encourage idiomatic and modular reuse of the Mathlib library.

• Collaborate with the global mathematics, Lean, and Mathlib open-source communities.

• Assist domain-specific formalization projects in areas like algebra, analysis, and topology.

• Create evaluation methodologies to benchmark machine-learning-driven formal reasoning tools.

⛳️ Requirements

• Ph.D. or Master’s degree in Mathematics, Computer Science, or a closely related quantitative discipline with a strong emphasis on formal methods, mathematical logic, or theoretical computer science.

• Strong mathematical foundation with the ability to comprehend, translate, and verify graduate-level mathematical proofs.

• Practical experience in writing formal proofs using Lean 4.

• Familiarity with the design and architecture of Mathlib.

• Solid software engineering principles in Python.

• Experience with LLMs, prompt engineering, and multi-agent developer tools.

• Capability to independently manage open-ended research and engineering projects in either a remote or collaborative environment.

• Preferred: Active contributor to Mathlib or other formal proof repositories like Coq or Isabelle/HOL.

• Preferred: Background in Machine Learning for Code, Automatic Theorem Proving, or reinforcement learning for symbolic reasoning.

• Preferred: Understanding of compiler design, AST manipulation, or parser development in relation to Lean.

• Preferred: Proven track record of open-source software contributions or research publications in formal methods, AI, or mathematics.

🏝️ Benefits

• Equal employment opportunity and non-discrimination policy.

• Inclusive work environment emphasizing diversity and career development.

• Opportunity to collaborate with global mathematics, Lean, and Mathlib open-source communities.

• Remote or collaborative working arrangements.

• Opportunities for professional and research collaboration.