Software Engineer

Confidential Search — Founding AI Engineer Seed-stage AI company backed by a top-tier venture firm About the company Our client is building an agentic service desk for managed service providers the firms that run IT for small and midsized businesses worldwide, a category representing roughly half a trillion dollars in annual spend. Their AI agents sit alongside human technicians: reading tickets, investigating root causes across the tools MSPs already run on, and executing approved fixes. The engineering problem is genuinely hard. These agents have to reason through problems no one wrote a runbook for, respect customer policy, carry state across workflows that run for hours or days, and behave predictably inside live production environments. The company closed a seed round from a tier-one venture firm and is now hiring the engineers who will define how the product actually works. The opportunity The Founding AI Engineer will own the agent layer end to end — from first prototype through hardened production deployment. The work rotates by nature: one week may go toward teaching an agent to diagnose an unfamiliar failure mode, the next toward building evaluation harnesses that surface where it breaks down, the next toward the tooling and infrastructure that let it act safely inside a customer's stack. This is a builder's seat, not an architect's. The person in this role will talk to customers directly, help set the roadmap, ship the code, and operate what they ship. They will also help settle the open questions in this category which decisions belong to agents, where humans keep the final call, how agent reasoning gets explained to a technician, and what a genuinely good technician experience looks like. Those standards are unwritten today. This role helps write them. What the role involves Building and refining AI agents that investigate IT issues, hold context over long sessions, call tools across customer systems, operate within policy, escalate for approval, and recover cleanly when something goes wrong. Standing up evaluation systems that quantify agent quality, surface failure modes, and convert production traces and customer feedback into concrete improvements. Running fast experiments across models, prompting strategies, retrieval, tool design, and agent architectures — then promoting what works into dependable production systems. Engineering the integrations and backend infrastructure that let agents run securely, reliably, and observably through long-lived customer workflows. Working directly with customers to understand their day-to-day, shipping product improvements end to end, and helping establish engineering practices as the team grows. Who this is a fit for Engineers who have shipped production software built on modern AI models: agents or other tool-using systems especially. People who move quickly with new models and techniques and can carry a promising prototype through to something reliable. Engineers comfortable evaluating nondeterministic systems, debugging failures without clean stack traces, and trading speed against quality, safety, and customer trust deliberately rather than by accident. Strong ownership, deep personal use of AI tooling, real product instincts, and comfort sitting in front of a customer and turning a vague problem into working software. Show more Show less