Lean 4 Proof Engineer - Mathematical Formalization
Alignerr • Remote
Education
Any
Type
Pay Rate
$185/task
Listed
Today
✅ Applying through this link gives you a verified candidate referral.
Referrals from verified candidates give your profile a visibility boost and help support our platform at no cost to you.
This position is hosted on an external talent platform. Please only apply for this position if it fits your skills and interests.
About this Role
What You'll Do
- Translate informal mathematical proofs into clean, structured, machine-verifiable Lean 4 formalizations
- Analyze proofs across domains — identifying gaps, hidden assumptions, and formalizable sub-structures
- Construct formalizations that test and extend the limits of existing proof assistants
- Investigate where automated provers break down and articulate the underlying reasons — complexity, missing lemmas, library gaps, and more
- Collaborate with researchers to design and refine formal verification strategies
- Develop highly readable, reproducible proof scripts aligned with mathematical best practices and proof assistant idioms
- Provide expert guidance on proof decomposition, lemma selection, and structuring techniques
- Formalize classical proofs and compare machine-verifiable structures against textbook arguments
- Create Lean proofs that reveal deeper patterns or generalizations implicit in the original mathematics
About the Role
What if your deep mathematical training could directly shape how AI understands and reasons about formal proof? We're looking for skilled mathematicians and formal verification specialists to translate advanced mathematical arguments into machine-verifiable Lean 4 proofs — working at the frontier of what proof assistants can express, capture, and automate. This is a fully remote, flexible contract role. If you live for rigorous proof construction and find satisfaction in the precision of formal systems, this is the role built for you.
- Organization: Alignerr
- Type: Hourly Contract
- Location: Remote
- Commitment: 10–40 hours/week
Who You Are
- Hold a Master's degree or higher in Mathematics, Logic, Theoretical Computer Science, or a closely related field
- Have a strong foundation in rigorous proof writing across areas such as algebra, analysis, topology, logic, or discrete mathematics
- Have hands-on experience with Lean (Lean 3 or Lean 4), Coq, Isabelle/HOL, Agda, or comparable proof systems — Lean 4 strongly preferred
- Deeply enthusiastic about formal verification, proof assistants, and the future of mechanized mathematics
- Able to translate informal arguments into clean, well-structured formal proofs with precision and clarity
Nice to Have
- Familiarity with type theory, the Curry-Howard correspondence, and proof automation tools
- Experience contributing to large-scale formalization projects such as Mathlib
- Exposure to theorem provers in contexts where automated reasoning frequently requires manual scaffolding
- Prior experience with data annotation, data quality, or AI evaluation systems
- Strong communication skills for explaining formalization decisions, edge cases, and reasoning strategies
Why Join Us
- Work on cutting-edge AI research projects alongside leading labs and research teams
- Fully remote and flexible — work when and where it suits you
- Freelance autonomy with intellectually stimulating, frontier-level work
- Contribute directly to how AI systems understand and reason about advanced mathematics
- Potential for ongoing work and contract extension as new projects launch
Requirements
- Fluent proficiency in English (Written & Verbal)
- Reliable high-speed internet connection
- Bachelor's degree or equivalent professional experience
- Demonstrated expertise in Mathematics
Compensation Analysis
What if your deep mathematical training could directly shape how AI understands and reasons about formal proof? We're looking for skilled mathematicians and formal verification specialists to translate advanced mathematical arguments into machine-verifiable Lean 4 proofs — working at the frontier of what proof assistants can express, capture, and a
Skills & Categories
Explore other opportunities in related specializations:
Related Jobs
Browse All Jobs from Alignerr
Discover more opportunities on Alignerr that match your skills and interests.
View All Alignerr Jobs →Community Reviews
Share your experience with Alignerr
Help other candidates make better decisions by leaving a review.
Sign in to leave a reviewLeave your review
Frequently Asked Questions
What is the assessment actually like?
Notoriously strict. Alignerr uses TestGorilla for role-specific timed tests — a blank coding environment for engineers, rigorous grammar and fact-checking for writers. There is almost no hand-holding. The critical catch: this is essentially a one-shot process. Fail or abandon the assessment, and you are typically locked out of that role permanently with no option to retake.
How quickly can I start earning after I pass?
Not immediately. Even after passing the assessment and completing identity verification (via Persona) and billing setup (via Deel), you may sit in a waiting pool for weeks or months. You only start earning when a project matching your specific skills launches and you are officially assigned. Do not plan around Alignerr income until you are actively on a project.
Is there a community?
Yes — and it is one of Alignerr's genuine strengths. Once assigned to a project, you are added to Slack channels where you can ask questions, get rubric clarifications from admins, and talk to other AI trainers. This is rare in AI training and makes a real difference when guidelines are ambiguous or change mid-project.
Is this traditional consulting?
Not exactly. You act as a "Teacher" for advanced AI. Instead of client deliverables, you are given complex scenarios to evaluate. You grade the AI's logic, correct its hallucinations, and provide expert-level reasoning. Your job is to train the model to think like you do.
Why is the pay so high?
This role requires deep, verified expertise. General knowledge isn't enough; the model is specifically being trained on "edge cases"—the rare, difficult, or highly technical nuances that only a senior professional would know.
What is the workload like?
This is cognitive, deep work. Unlike simple data labeling, you might spend 45-60 minutes on a single task, researching citations or verifying complex calculations. Quality is prioritized over speed.
What is the barrier to entry?
Alignerr is known for difficult technical assessments. You must pass a timed test in your specific domain (e.g., Python, Physics, or Language) before you are eligible for any paid projects.