Job Description & Details
This is a heavy-duty automation and multi-agent systems role where you will literally bridge the gap between LLMs like GPT, Claude, and Gemini and enterprise plumbing. If you enjoy moving past simple API wrappers and want to design robust, state-machine-backed workflows that actually survive production failures, this one is worth looking at.
What You'll Actually Be Doing
You will spend your days architecting the connective tissue for enterprise automation, tying together tools like Google Workspace, Box, and Figma with multi-agent orchestration frameworks. Expect to spend a lot of time designing durable state-management systems, handling idempotency, and building out human-in-the-loop approval loops so automated agents don't accidentally wreck production data while no one is watching.
The Core Tech Stack
You need deep, production-level chops in Python or TypeScript alongside modern orchestration tools like LangGraph, MCP, or A2A. The team won't hold your hand on REST APIs, OAuth flows, and complex JSON schemas, so you need to be entirely comfortable building structured data integrations from scratch and handling error recovery gracefully.
Interview Expectations
Expect the hiring team to hand you a scenario where an LLM agent goes into an infinite loop or drops a database connection mid-transaction, and they will want to see how you design state machines, retries, and idempotency keys to fix it. They are secretly checking if you understand the messy realities of distributed systems or if you only know how to build happy-path chatbot demos.
Application Advice
Make sure your resume doesn't just list the AI tools you've used, but specifically highlights your experience with workflow automation, state machines, and handling edge cases like retries and structured data integration. Emphasize any production deployments involving Python or TypeScript where you had to coordinate multiple services or APIs simultaneously.