Staff Software Engineer, AI Systems

Role overview Build the backend architecture powering agentic AI systems for an insurance underwriting platform that is AI-native by design. The role sits at the intersection of software engineering, data infrastructure, and applied AI, designing orchestration, tooling, and evaluation systems that bring LLM-driven workflows into production underwriting software.

Responsibilities - Design and implement agent workflows and orchestration systems for AI-powered product features. - Build backend services that integrate LLMs with structured insurance data and platform APIs. - Develop infrastructure for tooling, context management, and agent execution. - Contribute to internal frameworks supporting prompt iteration, evaluation, and observability. - Partner with product and design to translate underwriting workflows into AI-enabled product experiences. - Help define architectural patterns for building reliable AI-native features in production.

Requirements - 10+ years of experience in backend engineering, data engineering, or related roles. - Strong experience building production backend systems and APIs. - Professional experience working with LLMs, agent systems, or generative AI applications. - Hands-on experience with prompt design, tool-based agent architectures, or LLM workflows. - Proficiency in Python or a similar backend language. - Experience with cloud infrastructure and distributed systems, plus comfort working across layers of the stack.

Nice to have - Experience building production systems involving LLMs, agents, or prompt pipelines. - Experience designing internal platforms or developer tooling. - Familiarity with insurance, fintech, or B2B SaaS. - Contributions to open-source AI or data infrastructure projects. - Experience with TypeScript.

Benefits and work setup - Base salary range of $215,000 to $250,000 per year, with total compensation including stock options, benefits, and additional perks. - Final offer determined by candidate location, experience, and expertise.