Most backend roles touching “AI” mean a chatbot bolted onto a CRUD app. This one is different: the agentic workflows are the product surface.
You'll work on an established enterprise asset-and-metadata platform, building the AI engine that proposes metadata across the active project estate, the automated logic that reassigns or retires ownerless systems, and a live inventory of every agent and MCP server running across production and endpoint environments — all exposed through the platform API.
It's Go on AWS, against a real production codebase with real users. A large part of the job is reading code you didn't write, understanding how it behaves, and then extending, refactoring or decommissioning it without breaking anything.
What you'll own
AI-suggested metadata engine — proposing metadata across the active project estate, then tuning it until engineers actually accept the suggestions.
Automated ownership reassignment and deprecation — logic that finds ownerless or relinquished systems and acts on them safely.
Agent and MCP inventory — complete coverage across production and endpoint environments, accessible programmatically.
First-party product inventory and critical-user-journey mapping — making ownership and metadata trustworthy enough to decide on.
Legacy refactoring and decommissioning — with unit and integration tests covering everything you ship.
Sprint delivery alongside program and customer teams.
Requirements — check yourself against this list
You should be able to say yes to essentially all of these:
5+ years professional backend software engineering
Go as a working language — you've built and shipped scalable backend services and APIs in it
PostgreSQL — schema design and real querying
MongoDB — schema design and real querying
AWS — working knowledge of the services plus their client libraries and APIs
Agentic, AI-driven workflows you have designed, implemented and tuned — suggestion engines, automated decision logic, or assistant interfaces
Asset discovery / ingestion tooling — open-source or commercial (e.g. CloudQuery) for aggregating cloud resources and metadata
Data classification standards, dependency mapping, functional framework mapping
Navigating a large, unfamiliar codebase to refactor or decommission legacy code without regressing existing behaviour
Unit and integration testing for backend components, as standard practice