SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Sunset is a rapidly scaling business process automation platform that has grown from $0 to multi-eight-figure run rate in months, backed by top-tier investors including Floodgate and Afore. The company partners with frontier AI labs and enterprises to unlock proprietary business data as training material for next-generation AI models.
As Senior AI Product Engineer, you will own the end-to-end development of Sunset's dissolution support agent—a production LLM-backed product that helps businesses navigate complex shutdown and wind-down processes. This is a full-stack product engineering role, not research or infrastructure work.
You will design and ship retrieval systems, context selection, structured outputs, tool contracts, and orchestration logic that deliver correct, trustworthy customer outcomes. You'll move fluidly between customer experience design, application code, AI behavior tuning, evaluation frameworks, and production operations. Key responsibilities include:
- Owning the dissolution support agent from customer problem definition through production deployment, measurement, and iteration
- Designing grounded, inspectable AI behavior with explicit boundaries, permissions, and human-handoff experiences
- Building representative, versioned evaluations that measure routing accuracy, grounding quality, usefulness, safety, and real customer outcomes
- Shipping with production safeguards: permissions, tenant isolation, privacy controls, auditability, canaries, rollback, and recovery mechanisms
- Instrumenting runtime metrics: tool reliability, latency, cost, repeat contacts, support effort, and failure modes
- Determining which workflows should be fully automated, AI-assisted, human-structured, or deliberately human-owned—then building the approved approach
- Simplifying or removing agentic components when deterministic software would work better
- Collaborating closely with Product, Support, Security, domain experts, and full-stack engineers
Success means customers receive correct, useful resolution for meaningful dissolution needs with unsupported claims and unsafe actions contained within explicit guardrails. Human handoffs are timely and contextual. New intents move from evidence design through safe release using repeatable infrastructure. Quality, reliability, latency, cost, privacy, and customer effort remain visible as the product scales.
You should have personally owned production software products including LLM-backed capabilities beyond prototypes. You're a strong product engineer comfortable across customer experience, application code, backend systems, AI behavior, and production operations. You understand retrieval, context selection, structured outputs, tool use, orchestration, evaluation, and observability—and know when simpler deterministic software is better. You design for unsupported claims, uncertainty, permissions, privacy, escalation, rollback, and recovery. You use representative evidence to make ship/revise/stop decisions rather than optimizing demos. You learn domains quickly and work well with cross-functional partners. You use modern AI development tools fluently and verify their output with production rigor.
Bonus experience includes customer-support or workflow agents, LangGraph/LangChain orchestration, typed tool protocols, retrieval systems, golden datasets, offline evaluation, shadow deployments, model-based judges, and privacy-sensitive multi-tenant audited systems in legal or financial domains.
