We need a hungry, high-agency engineer with 1-4 years of experience building LLM-powered products at an early-stage B2B AI startup who has also sat with customers and owned deployments end to end. You should be comfortable writing production code most of the week and running a workshop with a chief underwriting officer in the room. Bonus points if you've worked on document ingestion, agentic workflows, or evaluation frameworks, and especially if you've been a founding or forward deployed engineer at a YC/Pear-grade company.
What you'll do
Discovery: Shadow underwriters on-site, map the life of a submission (actors, systems, handoffs, exceptions), and come back with a diagram, assumptions log, and a point of view on what to build first.
Build: Ship real integrations: LLM extraction and prompts tuned to each customer's documents, evaluation sets that prove it works on their data, underwriting guidelines turned into agent rules, intake from customer inboxes, and payloads into policy admin systems like Guidewire. You write production code in our platform alongside the core team, not scripts on the side.
Land: Run the pilot through go-live and stabilization. Train underwriters. Turn what you learn into product priorities and internal tooling that make the tenth deployment a fraction of the work of the first.
Lead weekly customer calls with some of the largest insurance companies in the US, showing integration progress and managing 8-12 week deployment cycles.
Build and maintain eval frameworks grounded in the scientific method: create ground truth datasets, measure recall and accuracy, and make data-driven decisions about what ships to production.
Collaborate with the core engineering team on backend and product features when integration load fluctuates, contributing across the full platform.
This is an on-site, direct-hire role in San Francisco, CA.
- 1-3 years of experience as a forward deployed engineer or founding engineer at an early-stage startup (pre-seed to Series A / pre-Series B, including YC or Pear-grade environments)
- Experience building LLM applications or working at AI-native startups
- Experience building agentic workflows (multi-tool call orchestration, agentic RAG, multi-agent workflows)
- Strong backend engineering skills shipping production services (Python, TypeScript, Postgres, SQL, Azure, or related full-stack work)
- Experience with enterprise B2B applications, document parsing, VLMs, and LLMs
- Scientific mindset for evaluating AI systems: ground truth datasets, recall/accuracy measurement, data-driven ship decisions
- Experience running customer integration calls and owning end-to-end technical outcomes
- Ability to work independently and with a team, comfortable with direct client interaction
- Willingness to work at high intensity with high responsiveness, including evenings and weekends as needed
- Ability to work in-person 5 days a week in San Francisco, CA (or relocate)
- Solid credentials and background in Computer Science
- Mission-driven, proactive, and eager to build