Lead AI Engineer/Architect with Mainframe/Cobol —Modernization

About

We are modernizing a large-scale enterprise insurance policy administration system from Visual Age Smalltalk / mainframe COBOL to Java 21 + Spring Boot + MongoDB. You will design and build the AI agents and skills that power every phase of the transformation — from reverse engineering and code extraction through to Java code generation and synthesis.

What You Will Do

• Build and maintain production AI agents for reverse engineering, code extraction, code transformation, synthesis and automated testing across Smalltalk and COBOL codebases
• Author reusable AI skills for repeatable subtasks (pseudocode conversion, diagram generation, business rule extraction, pattern identification)
• Design multi-agent pipelines with orchestration, audit logging, and human-in-the-loop checkpoints
• Own the agent standards library — design principles, naming conventions, and quality gates
• Mentor engineers on agentic AI patterns, prompt engineering, and legacy system analysis

Required Skills

• Hands-on experience building production AI agents (Anthropic, LangChain, LangGraph, AutoGen, or similar)
• Strong prompt engineering — system prompts, structured instructions, tool use, context window management
• 5+ years software engineering; 2+ years in agentic AI or LLM systems
• Proficiency in Java 21 and Python
• Ability to read and reason about legacy code (Smalltalk, COBOL, or equivalent)
• Experience with legacy-to-modern migration projects
• REST API design (OpenAPI 3.0) and MongoDB
• Git, CLI-first tooling, CI/CD pipelines

Nice to Have

• Insurance industry experience (policy administration, rating, underwriting)
• IMS/DB2, IBM MQ, or mainframe integration background