AI Engineer at Lumion Location: In-office, South Jordan, Utah Comp: Salary, meaningful equity, and bonuses tied to company revenue What Lumion does Lumion is the all-in-one operating system for trade and technical schools. We power the entire institution: AI-driven lead generation and conversion, enrollment, financial aid, student lifecycle management, and tuition payments. Most schools are still stitching together five or six disconnected tools to do what our platform handles. America has a massive skilled trades shortage and the schools that train electricians, welders, and nurses are the bottleneck. We're backed by TTV Capital and Tusk Venture Partners. Why this role matters At most companies, AI is a feature added onto an existing product. At Lumion it is the product. Our agents talk with people who are deciding whether to go back to school. They answer questions late on a Sunday night, follow up when a human wouldn't, and carry a prospective student from first question to enrolled. The reliability bar is high because the stakes are. These systems handle money, student records, and regulated communication. They cannot quote the wrong tuition, ignore an opt-out, or mislead a 19-year-old weighing a student loan. Making AI dependable under those conditions is one of the hardest problems in the field right now, and it is the problem you would own. The work You will own the systems behind Lumion's AI in production. Conversational agents. AI that talks with people over text, email, and live phone calls: qualifying a lead, answering questions, following up, and handing off to a human at the right moment. You will build the logic that decides when the agent reaches out, what it says, and when it escalates. Retrieval. Every school is different, so an agent has to answer from that specific school's programs, policies, and pricing rather than a generic guess. You will own the retrieval layer that keeps answers accurate and each customer's data cleanly separated. Multi-model infrastructure. Different tasks call for different models. You will own model selection, cost, and latency, using cheaper models where they hold up and the expensive ones only where they pay off. Voice. Real-time phone conversations bring their own problems: response timing, interruptions, hold handling, and prompts that survive a caller talking over the agent. Your job is to make these calls hard to distinguish from a person. Reliability. Evaluations, guardrails, and monitoring. This is the engineering discipline that makes a prototype trustworthy enough to run a school's enrollment pipeline. How we build You need to be fluent with agentic coding tools like Claude Code, Codex, or Cursor. We care about how you work, not which tool you use. If your AI experience stops at pasting prompts into ChatGPT, this is not the right role. The tools also do not replace judgment. You need to recognize when the output is wrong, and when a quick fix will cost you later. This is product engineering. You ship to customers and iterate fast, and whatever you build has to hold up in production from day one. This is an in-office role in South Jordan, Utah, by design. Hard problems get solved faster in person, and we want whoever owns our AI working directly alongside the team and the CTO. Who you are You have shipped production AI. Real systems that users relied on, where you dealt with hallucination, runaway costs, latency limits, and the failures that only show up at scale. You have strong fundamentals. Data modeling, API design, system architecture, and the ability to debug a production incident under pressure. The AI sits on top of a serious platform, so you need to be solid on both. You have product sense. You understand why a requirement exists, not only how to build it. That matters most when money, student data, and compliance are involved. You're sharp, not credentialed. Your degree and years of experience don't matter to us. What you can build and how fast you learn do. Why you'll want this Ownership. You will own Lumion's entire AI surface. At a larger company you might own a single prompt and wait weeks to ship. Here you own the agents, retrieval, voice, and reliability, and you ship constantly. Equity. Your stock is tied to revenue that is growing, so what you build connects directly to what your equity is worth. Impact. The people on the other end of this software go on to become electricians, nurses, and welders. The work matters, and you will feel it. Speed. Strong product direction, an AI-native engineering culture, and the autonomy to move quickly without layers of process. Comp and benefits Salary based on what you can do Meaningful stock options Bonuses tied to company revenue MacBook Pro and the AI tooling you need Unlimited PTO 401(k) Health, dental, and vision reimbursement up to $537.50/mo $200/mo tech stipend $100/mo wellness stipend Must be authorized to work in the US. We do not sponsor work visas. This is an in-office role in South Jordan, Utah, and relocation support is available for the right person. How to apply Skip the cover letter. Send us the best AI system you have shipped, what it did in production, and what went wrong along the way. We will get into the prompt engineering and system design when we talk.