AI Engineer

<div class="content-intro"><p>ON.energy<span class="Apple-converted-space">&nbsp;</span>is setting the standard for large load interconnection. ON’s patented AI UPS™ is a medium-voltage UPS that serves as a firewall, protecting the data center from the grid, and the grid from the data center.&nbsp;With multiple gigawatts currently under construction, we are enabling grid-safe data centers.&nbsp;</p></div><h3 class="elementToProof" data-ogsc="rgb(0, 0, 0)">Role Summary</h3> <div class="elementToProof" data-ogsc="rgb(0, 0, 0)">This is the first dedicated AI engineering role at<span class="Apple-converted-space">&nbsp;</span>ON.energy. There is no established AI platform team to inherit and no fully scoped roadmap waiting for you; you will help decide what gets built, what gets bought, and what patterns the rest of the company adopts. If you want a role where the early decisions are still open, this is one. If you need a mature platform and a defined backlog on day one, it is not.</div> <div class="elementToProof" data-ogsc="rgb(0, 0, 0)">You will report to the Program Director, AI Enablement, who owns intake and prioritization of AI use cases from across the business. This is a small, direct working relationship where you will be the technical half of it. The role is based in Houston, with a minimum of three days per week in the office and a reasonable remote setup for the days you are not. Occasional travel should be expected.</div> <div class="elementToProof"> <div class="elementToProof" data-ogsc="rgb(0, 0, 0)">&nbsp;</div> <h3 class="elementToProof" data-ogsc="rgb(0, 0, 0)">Key Responsibilities</h3> </div> <ul> <li data-ogsc="rgb(0, 0, 0)">Building production AI systems end to end, from prototype through deploy and monitor — owning the delivery dates and milestones you commit to, and flagging risks and contingencies early rather than reporting them after the fact</li> <li data-ogsc="rgb(0, 0, 0)">Working directly with stakeholders across Engineering, Operations, HR, Finance, and Commercial to translate workflow problems into shippable tools, much of it in person and side by side with the teams whose work you are changing</li> <li data-ogsc="rgb(0, 0, 0)">Writing the evaluation harnesses, prompts, retrieval logic, and integration code that make LLM-powered features reliable</li> <li data-ogsc="rgb(0, 0, 0)">Establishing reusable technical patterns — prompt and model governance, eval frameworks, data access, deploy practices — built in partnership with IT so they fit our security, identity, and infrastructure standards</li> <li data-ogsc="rgb(0, 0, 0)">Partnering