Hunt down high-leverage processes. Talk to people across the company, find the manual, repetitive, error-prone work, and rank it by what's genuinely worth automating.
Pull a workflow out of someone's head and specify it. Most of the job is turning a vague "this is just how we do it" into something precise enough to automate.
Build the automation. Python and API glue, MCP servers and clients, Claude Skills, agent workflows, and small services where they're needed.
Instrument what you ship so it doesn't rot silently. Logs, alerts when something breaks, and a way to know whether the automation is still doing its job.
Drive adoption. Get non-technical teams to trust and use what you've built, train them, and stay close after launch so it sticks.
Know when to escalate. When an automation needs to become real platform, hand it to the backend team instead of bolting on. Knowing where that line sits is part of the job.

You've automated real business processes end to end, from "what's the actual problem here" through to people using it every day. We'll want to see what you shipped and what it saved.
You're comfortable in front of non-technical teams. You can extract a workflow, push back on a bad priority, and sell a new way of working to people who don't code.
You write code. Python, APIs, scripting, enough to build the automation yourself without waiting on an engineer for every step.
You're fluent in the agentic toolchain (Claude Cowork, Skills, MCP, workflow tools), or you'll obviously get there fast.
You instrument and maintain what you build. You've felt the pain of an automation that broke quietly, and you don't let it happen twice.

AI Automation Engineer

Key skills