COBOL Programming Expert

About the Opportunity

A leading AI research organization is evaluating how advanced AI systems perform in specialized engineering domains, and is seeking expert engineers with deep, hands-on production experience in COBOL. The immediate focus is a benchmark of roughly 100 tasks centered on database migration in COBOL — you'll apply your mainframe expertise to assess complex, real-world technical scenarios and directly shape how cutting-edge AI performs on mainframe systems.

What You'll Do

• Apply your language expertise to evaluate technical tasks against real-world professional standards
• Review intricate code-level situations — including database migration scenarios — and provide precise, structured written assessments
• Provide clear written rationales explaining your expert judgments
• Complete well-defined, time-bounded tasks with explicit evaluation criteria

Who We're Looking For

We value strong engineering fundamentals, fast ramp-up, and high ownership. The ideal candidate brings:

• 5+ years of professional COBOL development — IBM mainframe (z/OS) production work; JCL, CICS, VSAM, DB2; banking, insurance, or government systems
• Demonstrated project, professional, or research experience in one of the following supplemental languages: Rust, C++, Java, Kotlin/JVM, Python
• Experience with financial system backends (e.g., banking, insurance) is a strong plus, especially anything touching data or database migration work
• Ability to articulate not just what code does but why it's correct or idiomatic, clearly in writing
• Welcomed: practitioners from banking, insurance, or government backgrounds; non-traditional digital footprints are fine — a GitHub profile is not required

Why This Work Matters

COBOL runs critical infrastructure in global banking, insurance, and government systems, yet remains among the least-represented domains in AI research. The expertise you bring is rare, and your assessments directly influence how AI systems learn to operate in it.

Engagement Details

• Compensation: $100–200/hour, based on depth and experience
• This is a one-week sprint: a 2–3 day pilot, iteration on the production task set, then roughly a week of production work. Kickoff is targeted for this Friday, and we need experts who can stay highly engaged throughout
• No proprietary tooling required — tasks are completable without employer-provided systems (no mainframe access needed)