Machine Learning Engineer

Most AI products are a thin layer of prompts around a model someone else trained. The hard parts, retrieval that stays accurate over messy documents, conversation state that holds without drifting or hallucinating, agents that reliably execute multi-step work, get treated as a feature bolted onto something bigger. Here they are the product. An early-stage company being incubated by a T1 VC is rebuilding a category of enterprise software from scratch, AI-native rather than retrofitted onto a legacy system. The founders ran platform engineering at enterprise scale before starting this, so the credibility and the customer pull are already there. This is a foundational hire. You would own the applied AI layer end to end: Retrieval : high-accuracy, attributable RAG over domain-specific document sets where the data is genuinely messy. Chunking, embedding strategy, and retrieval precision are real problems that need to be solved here. Conversational systems : proper state and context management, intent handling, and orchestration over backend models. Agents : workflows that execute, and not just answer questions. All of it production-grade, in front of users, reliable under real load. What the role needs: you have shipped production AI systems that real people used at scale, and you can point to what you built, who used it, and how it held up. You have depth in RAG and retrieval or in conversational and agentic systems, ideally both. Breadth and curiosity beat fifteen years on one narrow problem. You define ambiguous problems and drive them rather than waiting for a spec, and you already build with AI-assisted tooling. Where: Bay Area, roughly four days a week onsite. On offer: base between $240k and $300k, meaningful founding-level equity, and a say in the architecture and the team from day one. If you are tired of AI work that lives as a feature inside someone else's roadmap and want it to be the thing customers depend on, this is the seat. Show more Show less