Software Engineer - AI Platform

Job description

About Us We are building an agentic mesh network for US healthcare, deploying multimodal systems that can autonomously handle non-clinical tasks and with seamless execution between entities. Our founding team is ex-Palantir with direct experience delivering mission-critical software to HHS, NIH, and other federal institutions. We are based in New York City and work in-office every day. We operate with an extremely flat structure. We ship, we learn, we compound. The Role You will design and build core systems that power autonomous agent orchestration across the most critical vertical within our country. This is not a feature-team role. You will own real subsystems with the autonomy and accountability that most engineers don't see until years into their careers. What You Will Build - Agent runtime — execution engine, task scheduling, state management, exception handling, and human-escalation pathways - Complex agent orchestration — multi-agent coordination, tool use, real-time decision-making pipelines that process clinical documents, voice interactions, and multimodal healthcare data - Federated interoperability protocols that enforce data residency while enabling cross-entity agent collaboration - Developer-facing abstractions — APIs, SDKs, and workflow primitives that allow both internal and external teams to deploy agents on the platform Who You Are - 2+ years building production software. You've shipped real systems — not just side projects, not demos. You've debugged something at 2am and know what it means to own what you build. Yes, Claude Code does a lot of the work, but you deeply understanding how to get compounding leverage from it and other tools every day. - You're excited by distributed agent systems — state machines, message guarantees, fault tolerance, consensus. You may not have designed these from scratch yet, but you've thought hard about them and want to. - You're drawn to pushing the envelope of Applied AI. You want to put the latest models into production — not just play with them. You understand that the gap between a demo and a deployed system is where the real engineering lives. - You operate with extreme ownership. You don't wait for specs. You identify the highest-leverage problem and solve it. You'd rather ask forgiveness than wait for permission. - You're drawn to a sense of mission and purpose. You want your work to matter — not in the abstract, but in a way you can point to. Compensation Top-of-market cash compensation plus significant founding-team equity. We pay for the best because we hire the best. New York City. In-office.