Senior Lead Software Engineer, AI Systems- 11509

What You'll Do:

• Design and build the backend services behind the agent platform — the tool layer over Coupa APIs, context/state management, and the execution environment that keeps supplier data isolated.

• Build the evaluation and logging pipeline: how we measure behavior, catch regressions, and turn production failures into fixed behaviors.

• Partner with product and engineering to ship the first supplier-facing use cases (invoicing, catalog, strategic views) on the platform you build.

• Own how the platform reaches production — deployment, configuration, scaling, and staying secure.

• Make architectural calls on runtime isolation, credential handling, audit logging, and tenant separation.

• Raise the engineering bar for the team — code quality, testing, and operational discipline.

What You Will Bring to Coupa:

Have 6+ years shipping production software and are strong in Python (TypeScript a plus).

Have built and operated services on a major cloud (AWS preferred) — containerized deployments, IaC, and the day-two work of keeping them running.

Have built backend services, platforms, or infrastructure that other engineers or systems depend on — APIs, SDKs, runtimes, internal platforms.

Have operated what you've built in production: you understand reliability, observability, failure modes, and least-privilege security, not just the application layer.

Write clean, well-tested code and reason about latency, failure modes, and user outcomes together, not as separate concerns.

Write and speak English fluently — the team operates in English and the work is prose-heavy.

Are comfortable working near LLMs — you don't need to have shipped an AI agent, but you should be curious about the space and quick to pick it up. Familiarity with LangChain / LangGraph or similar agent frameworks is a plus.

Bonus Points

A public GitHub, blog, or portfolio we can look at (strongly preferred — it's the fastest way for us to see how you think).

Are a daily user of agentic coding tools (Claude Code, Cursor, Codex, or equivalents).

Have shipped an LLM-powered feature or product, and have opinions about why most of them don't work.

Experience with agent frameworks, eval tooling, or production LLM observability.

Hands-on with our specific stack — AWS Bedrock / AgentCore, LangGraph, or comparable managed-LLM and sandboxed-execution infrastructure.

Experience in procurement, fintech, or another regulated enterprise domain.