- 8+ years building production AI systems, with the past few years deeply engaged in LLM-native development: agents, evals, retrieval, structured generation, and human-in-the-loop design over complex document workflows. - Strong instincts on the generative side: narrative drafting, evidence synthesis, and structured argumentation, not just extraction and classification. You see generative quality as where real margin lives. - Strong data modeling instincts: schema design and evolution, provenance tracking, and conflict resolution at scale. - Track record of translating messy, high-stakes operational workflows into reliable AI systems, ideally in regulated, document-heavy domains (legal, healthcare, compliance, financial services). - You've built confidence and quality frameworks for ambiguity: scoring, exception routing, and feedback loops that make systems more trustworthy over time rather than accumulating tech debt. - Strong product and systems judgment: you know when to ship a feature, when to invest in the platform, and how to sequence the two. - You've recruited, hired, and led strong AI/ML engineers, and you've stayed close to the technical work. - Thrives in a fast-moving, high-candor, in-office environment.