Description
Summary Our support team is drowning in repetitive requests that a smart agent could handle end-to-end - but only if it can actually interact with our systems, not just talk. We're after someone who can build an AI agent that reads a customer's request, decides what needs to happen, calls the right internal tool to make it happen, and pulls in a human the moment something is beyond its remit. Plain Q&A chatbots aren't what we need here. How it should behave: - Reach into our internal tools to fetch order info, check statuses, and log support tickets - Work out which tool fits the request and call it at the right moment - Escalate cleanly to a human whenever the matter is sensitive or outside its defined scope - Never guess an action or operate on data that's missing or wrong - reliability is non-negotiable What you'd deliver: - A LangChain agent built around a tool-calling architecture - Integration of 3-4 internal APIs as tools (order lookup, status check, ticket creation) - A validation and guardrail layer so actions only fire when they're safe and intended - A straightforward escalation route to human support for edge cases - Stable, predictable operation in a production setting Ideal candidate: - Has genuinely shipped tool-calling agents before, beyond simple chatbots - Strong hands-on skills in LangChain and Python - Experienced connecting agents to internal APIs and tools - Understands guardrails, validation, and how to keep an agent predictable - Delivers clean, well-documented code Would be great if you also have: - Prior work on human-in-the-loop handoff logic - Experience in customer-support or service automation If this is your wheelhouse, drop a relevant example or portfolio link when you apply. We are looking to get started right away.