About the Role
AI Agent BuilderNational Technology Management (NTM) | Bingham Farms, Michigan | Onsite | Full-TimeThe short versionWe need someone who has already put an AI agent into production, watched it fail in a way they did not expect, fixed it, and then proved with numbers that it was working.NTM delivers AI-driven Managed IT and cybersecurity. The AI layer here is not a pilot. We have agents in production today handling inbound calls, intake and triage, lead and pipeline work, client-facing chat, and internal ticket routing across our client base. They talk to our core business systems and people notice when they go down.You are joining a running system, not starting one.What you will buildAgents and automations that permanently remove manual work, for us and for our clients. Some of what is queued:
• An executive assistant agent that runs the administrative layer of a C-level day: email triage, scheduling, prioritization, follow-ups, meeting prep
• Scheduling and intake agents that take the back and forth out of booking and qualifying
• Website and campaign build workflows that hand off cleanly into our CRM platform
• Lead generation systems with dashboards that show what actually converted
• Multi-agent workflows that retire repetitive internal processes, with the reporting to prove itYou will get the problem and the outcome we want. How it gets built is yours.What we requireYou build on top of existing LLMs. You are not training models. You are engineering around them: prompt architecture, tool and function calling, retrieval, structured output, state across turns, and picking the right model for the job instead of defaulting to the most expensive one.You have shipped to production. Not a demo. Something that ran unattended, wrote to a real system of record, and had users.You measure your own work. You know your agent's cost per run, its failure modes, and what it does when a tool call returns garbage. You can tell the difference between a prompt change that helped and one that just changed the output. If you have never had to answer for the accuracy or the bill of something you deployed, you are early for this seat.You move without a finished spec. You get a problem and a business outcome, not acceptance criteria. You come back with something running.Tech stackAreaToolsPlatformsGoHighLevel, OpenClawLanguagesTypeScript, Python, ReactAIClaude (Anthropic), OpenAI APIsAutomationn8n, Rewst, Power AutomateDeploymentRailway, Cloudflare, GitHub ActionsDatabasePostgreSQLYou do not need all of it on day one. You do need to pick up the gaps inside a week.CompensationBase salary plus company performance bonus, up to 25% of salary depending on where the company lands. Everyone here participates. We are packaging these automations with our IT services to grow aggressively this year, and the people who build the systems share in what they produce.LocationOnsite in Bingham Farms, Michigan, working next to the people whose problems you are solving. This role does not work remotely.How to applySend your GitHub and one thing you built that ran in production. Not a resume bullet. Tell us what the problem was, what you built, and how you knew it worked.That last answer is the one we read closely.We review everything and we decide fast.trustntm.com#NTM #AIJobs #AIAgents #Automation #MichiganTech