Python Automation Engineer (AI-Driven)

Our Commitment to a Human-First Hiring Process
Because we deeply value human connection, our recruitment process does not involve automated AI screening tools or AI agents. From the moment you submit your application to the final conversation, every resume review, hiring decision, and message comes directly from a real person on our team. We believe meaningful collaboration starts with genuine human interaction, and we treat your time, background, and effort with the personal attention they deserve.

About the role
Our customer’s(USA) AI Augmentation team builds AI tools that take real work off people across the company, in HR, legal, marketing, finance and operations. Six solutions run in production today. The next step is a shared layer underneath them, so that every new solution starts with authentication, configuration, logging, model access and our hard-won lessons already in place, and so that people beyond the core team can build and ship safely.
You’ll take a hands-on engineering role on that shared layer. You will write much of the code yourself, working alongside AI coding agents every day, set the engineering bar others follow, and make the technical calls that let the program scale. You report to the VP, AI Augmentation.

What you'll do
• Build and own the shared Python library every solution imports for authentication, configuration, logging and request context, and move the existing solutions onto it.
• Build the project template a new solution starts from, so a new builder goes from first commit to first production deployment in days rather than weeks.
• Design and build shared services, starting with an AI gateway that gives every solution model access, cost per solution and one audit log, with de-identification through the company's approved service.
• Settle model, framework and library choices with measured bake-offs on real tasks, and record each decision, its evidence and its fallbacks in the repo.
• Own the internal platforms already running: a skills marketplace for AI coding agents, an architecture and engineering-assessment tool, an AI tools portal, and a demo hub whose generator records narrated videos of apps working.
• Ship on AWS EKS through GitLab CI: pipelines, versioned releases, configuration and secrets through the platform team's standards, and gates that fail closed.
• Assess solutions, new and inherited, for production readiness across security, reliability and testing, and turn the gaps into a plan with an owner and a date.
• Mentor developers growing into the platform, review their work, and write the docs and agent skills that let someone pick up a piece without you in the room.

What you bring
• [10+] years building and running production software, including several years at a very high level setting technical direction for other engineers.
• Strong Python, including FastAPI services and a packaged library that other teams depend on. Working Node.js and TypeScript.
• Daily, hands-on use of AI coding agents (Claude Code or similar) to ship production code, and a clear view of where they go wrong and how you check their work.
• Production experience with LLM APIs: prompt and output design, evaluation against real tasks, cost and latency tradeoffs, and fallback when a model is unavailable or not approved.
• AWS, Kubernetes (EKS) and containers, and CI/CD pipelines you have written yourself, ideally GitLab CI.
• Authentication and authorization in practice: OIDC and OAuth2 with an identity provider such as Okta, per-user access to a person's own data, least privilege, and secrets kept out of code, logs and chat.
• A testing habit that proves behavior: browser tests (Playwright or similar) that click real controls against a real server, unit and integration tests, and CI gates that cannot pass without running.
• Clear, plain writing: decision records, status that says exactly what is built, verified and signed off, and requests a busy executive can answer in one line.
• Exceptional verbal and written English communication. You must be able to explain technical automation logic simply and clearly to non-technical stakeholders in Legal, HR, or Operations, building trust and enthusiasm for your solutions.
• Being comfortable working later hours (e.g., 2:00 PM – 7:00 PM or 4:00 PM – 8:00 PM CET) to facilitate real-time collaboration, stand-ups, and rapid feedback loops with US-based counterparts.

Nice to have
• Healthcare or other regulated data: HIPAA, PHI handling, de-identification and audit logging.
• AWS Bedrock, a model gateway such as LiteLLM, or AWS AI services such as Polly.
• Next.js or React front ends, and Postgres with schema migrations.
• An internal developer platform, template or shared library that other teams adopted, with numbers to show for it.
• Architecture modeling (C4, Structurizr) and assessment against published practice such as OWASP ASVS, the OWASP LLM Top 10 or AWS Well-Architected.

How we work
•

AI agents are teammates. Each solution has its own agent working in its own repo under shared rules. You will direct agents, review what they produce, and improve the rules and skills they follow.
• Decisions come from evidence. A real choice between models, frameworks or approaches gets a bake-off, and the decision record cites it.
• A control is not tested until a test clicks it. Tests run against the real page and the real server, and a check that can report success without doing its work is treated as a bug.
• Status is plain and honest. Built, live on nonprod, verified end to end, or signed off by the person who asked. Never one rounded up to the next.
• Security first. Least privilege, per-user access, no PHI outside approved systems, and Security review before anything new reaches users.
• Small, visible changes. One commit per fix or feature, released with versioned tags, each shipped with a short demo.
Engagement
• Type: B2B Contract
• Reports to: Customer’s VP of AI Augmentation.
• Hours: 40/week
• Location and time zone: remote with significant daily overlap with the US Eastern Time zone. 
• Start date: ASAP