Job Description & Details
This is a heavy-hitting infrastructure and integration role focused on building the connective tissue between enterprise Microsoft 365 data and modern LLMs like ChatGPT and Claude. You'll be architecting Model Context Protocol (MCP) services and AWS runtimes to safely expose OneDrive and SharePoint content to AI agents without creating massive security holes. If you enjoy solving gnarly auth problems and pushing the boundaries of agentic workflows in production, this is a phenomenal technical challenge.
What You'll Actually Be Doing
Expect to spend your days designing and running robust gateway services on AWS that handle identity, session state, and observability for AI agents. You will write code to bridge Microsoft Graph APIs with LLM tool-calling architectures, dealing with the messy realities of OAuth 2.0 flows, delegated consent, and SharePoint permissions. A lot of your time will go into ensuring that when Claude or ChatGPT interacts with enterprise documents, it does so securely, efficiently, and with zero leakage of sensitive corporate data.
The Core Tech Stack
AWS is your bread and butter here—you need deep fluency in IAM, networking, serverless or containers, and secrets management to keep these agent runtimes locked down. On top of that, you must have hands-on production chops with Microsoft 365 integrations, specifically Microsoft Graph, Entra ID, and SharePoint site-scoped access. Finally, you should already know your way around OpenAI and Anthropic developer platforms, particularly how context windows, tool calling, and LLM output limitations actually behave in the wild.
Interview Expectations
When you sit down with the engineering team, expect them to grill you on failure modes in agentic systems, such as how you would handle an infinite tool-calling loop or token exhaustion mid-session. They are secretly testing whether you understand distributed systems failure patterns or if you are just blindly trusting LLM outputs. You will also likely be asked to whiteboard a secure authentication flow using OAuth 2.0 and Microsoft Entra ID to show how you prevent unauthorized data exfiltration between an LLM and corporate SharePoint repositories.
Application Advice
Do not just list AWS and Python on your resume and hope for the best. You need to explicitly highlight production instances where you secured APIs using OAuth 2.0 or integrated enterprise identity providers like Entra ID. Make sure keywords like Model Context Protocol, Microsoft Graph, and Amazon Bedrock are front and center, as the ATS will definitely be filtering for these exact integration domains. Tell a clear story about how you've scaled cloud infrastructure and managed thorny permission models in past enterprise roles.