Agentic Developer, LLM‑Native / Enterprise (Contractor), West Coast/ NYC

What You’ll Do

• Generate, refactor, and extend enterprise‑grade codebases by orchestrating LLMs as first‑class development tools, not just assistants
• Modernize and evolve legacy systems (often written in traditional enterprise stacks) using AI‑driven development workflows
• Apply “vibe coding” practices responsibly within highly regulated domains such as financial services, capital markets, or other compliance‑heavy environments
• Design and guide agentic workflows where LLMs reason, iterate, and produce code aligned with business and architectural constraints
• Validate, test, and harden AI‑generated code to meet production, security, and audit standards
• Collaborate closely with product managers, architects, and clients to translate ambiguous requirements into working systems - fast
• Influence how Lab49 and its clients adopt LLM‑native software engineering practices at scale

What We’re Looking For

• Strong software engineering background with experience in enterprise or legacy systems
• Hands-on experience manufacturing production-grade software in AI-assisted way at scale (not just autocomplete or chat usage, but sustained code production through prompts, agents, or workflows)
• Experience working within mature CI/CD environments, with automated testing fully integrated into the delivery pipeline
• Comfort working in regulated, risk‑aware environments, where correctness, explainability, and traceability matter
• Ability to reason about system behavior, edge cases, and failure modes- even when the code is AI‑generated
• Fluency in at least one major enterprise language or ecosystem (e.g., Java, C#, Python, JVM‑based stacks, etc.)
• Strong intuition for when to trust the model- and when not to
• Leading teams through the AI adoption journey and helping enterprises build scalable, robust AI software factory frameworks

Nice to Have

• Experience designing or using agentic development workflows (multi‑step prompting, tool‑using agents, code‑generation loops)…