Job Description & Details
This is a heavy-duty production AI engineering role built around deploying actual LLM-powered agents to help field technicians in the real world. If you are tired of building endless AI demos and proofs-of-concept that never see the light of day, this role actually puts complex multi-agent systems into live operational environments.
What You'll Actually Be Doing
You will spend your days designing, building, and hardening multi-agent orchestration pipelines that rely on conditional routing, state management, and real error recovery instead of hoping the model just works. You'll architect RAG pipelines featuring hybrid search, re-ranking, and Reciprocal Rank Fusion, while also wiring up tool-calling agents that execute parallel APIs. Expect to tackle real-time streaming via WebSockets or Server-Sent Events, integrate Azure OpenAI endpoints with robust circuit breakers, and handle multi-modal vision inputs for image analysis.
The Core Tech Stack
You need absolute mastery of modern Python and async programming—specifically FastAPI, asyncio, and httpx—because blocking event loops will immediately kill the real-time streaming requirements here. On the AI side, deep hands-on experience with LangChain, LangGraph, and Azure OpenAI is non-negotiable. You'll also need solid database fundamentals with PostgreSQL and async SQLAlchemy, paired with containerization and microservices architecture to keep everything running under production load.
Interview Expectations
The engineering panel will likely ask you to design a resilient multi-agent architecture that handles downstream API failures gracefully without dropping the user's conversational state. They want to see how you implement circuit breakers, fallback chains, and state management using LangGraph. Another heavy focus will be on optimizing your RAG pipelines; expect questions on how you tune hybrid search, manage chunking strategies, and implement Reciprocal Rank Fusion to reduce hallucinations in a high-stakes operational setting.
Application Advice
Skip the generic AI buzzwords on your resume and focus strictly on your production Python and LLM credentials. Make sure keywords like LangGraph, async Python, FastAPI, vector search, and circuit breaker patterns are explicitly highlighted in your experience bullets. The hiring managers want proof that you know how to build resilient, observable AI software that survives real-world chaos, so emphasize systems reliability and monitoring just as much as your prompt engineering skills.