SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Jerry is a YC-backed, profitable Series C startup with 5M+ customers building an AI-powered advisor to manage physical assets (cars, homes, motorcycles, RVs, boats). The company has pioneered its own infrastructure to automate sales and service conversations, currently handling 70% of inbound requests—over 50,000 chats monthly in production.
As Senior AI Systems Architect, you'll own the AI systems layer of this consumer product, working directly with OpenAI to integrate Jerry services into ChatGPT and pioneer new model capabilities. This is a high-impact individual contributor role with direct access to founders and executives.
Key responsibilities include:
- Defining the AI roadmap and making investment decisions on emerging technologies (voicebot, computer use agents, workflow automation)
- Designing evaluation frameworks and guardrails, balancing latency, cost, and accuracy in a regulated production environment
- Setting technical standards for chatbot systems and guiding the prompt engineering team
- Building new automations across record disputes, email/callout reduction, and document validation
- Partnering with subject matter experts to translate domain knowledge into production systems
- Evaluating and deploying next-generation voice models and workflow automation tools
You'll tackle the hard problems beyond prompting: agent permissions, system architecture, guardrails, and determinism in AI outputs. The role requires someone who thinks in interfaces, failure modes, and blast radius—someone who redesigns messy systems rather than working around them.
Ideal candidates have 5+ years in product or forward-deployed engineering (preferably early-stage startup experience), a track record of turning ambiguous problems into scalable systems, and hands-on production experience with LLM-based systems including evals, regressions, latency optimization, cost management, and failure handling. You should be fluent in modern AI, read release notes when new models drop, and have built with them. Pragmatism is essential—knowing when to build durably versus shipping what works this quarter.
