Mathematical Formalization Specialist

Formal mathematics

Translate advanced human-written arguments into formal proofs that a machine can verify. The work explores the limits of proof assistants and contributes to research on AI reasoning and reliability.

• Produce clear, correct formalizations in Lean or related proof systems.
• Analyze arguments for omitted assumptions, gaps, and components that can be formalized separately.
• Develop proofs for difficult cases where existing automation is insufficient.
• Help researchers refine methods and workflows for formal verification.
• Write proof scripts that are readable, reproducible, and mathematically well structured.
• Advise on decomposition, suitable lemmas, and the organization of formal arguments.
• Explain failures of automated proving, including complexity, missing results, and inadequate libraries.

Mathematical background

A master's degree or higher in Mathematics, Logic, Theoretical Computer Science, or a closely related subject is required. You need fluency in rigorous proof construction across areas such as algebra, analysis, topology, logic, and discrete mathematics.

Hands-on proof assistant experience is essential. Lean 3 or Lean 4 is strongly preferred, while Coq, Isabelle/HOL, or Agda experience is relevant; you should be able to convert informal reasoning into structured, verifiable proofs without extensive guidance.

Useful experience

Knowledge of type theory, the Curry–Howard correspondence, and proof automation is helpful. Contributions to large formalization efforts such as mathlib and experience with problems requiring manual intervention are valuable.

Strong communication also helps when explaining proof strategies, formalization choices, and difficult cases to collaborators.

Examples and arrangement

Assignments may involve formalizing a classical theorem and comparing its verified structure with the textbook argument. Other work includes documenting the exact point where automated reasoning fails or constructing Lean proofs that reveal implicit patterns and generalizations.

The engagement is an hourly contract with fully remote, asynchronous work and a flexible commitment. Contributors retain scheduling autonomy while working with AI researchers, and later projects may support continued work or an extension.