Software Architect - 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 define the architecture of the systems that power autonomous agent orchestration across the most critical vertical within our country. This is not a feature-team role. You will own entire subsystems — from protocol design to production deployment — with the autonomy and accountability of a technical co-founder. What You Will Build - Agent runtime architecture — designing the execution engine, task scheduling, state management, exception handling, and human-escalation pathways that the entire platform runs on - Multi-agent orchestration layer — the coordination primitives, tool-use frameworks, and real-time decision-making pipelines that process clinical documents, voice interactions, and multimodal healthcare data at institutional scale - Federated interoperability protocols that enforce data residency while enabling cross-entity agent collaboration — you will define the boundaries of what moves, what doesn't, and how agents transact across trust domains - Platform abstractions — APIs, SDKs, and workflow primitives that allow both internal and external teams to deploy agents on the platform. You will decide what gets exposed, what gets hidden, and where the right seams are. Who You Are - 5+ years building production distributed systems. You have designed, shipped, and operated infrastructure that runs at institutional scale. You've made architectural bets that held up — and learned from the ones that didn't. - You think in systems, not features. State machines, consensus protocols, fault domains, message guarantees — these are your native vocabulary. You see the failure modes before the happy path is built. - You're drawn to pushing the envelope of Applied AI. You want to architect the systems that put frontier models into production — not just integrate them. You understand that the gap between a demo and a deployed system is where the real engineering lives, and you want to define the patterns that make it repeatable. - 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.