Relari

AI Agent Builder for Software 3.0

Hiring — 1 openYC-W24Early

What Relari does

Relari is building the infrastructure for trustworthy AI agents in the Software 3.0 era. We’re the company behind Nuvi.dev — a natural language agent builder that turns plain-English specifications into intelligent, verifiable agents — as well as the open-source frameworks Agent Contract and continuous-eval, which ensure AI systems behave as intended.

1 open role

Founding Engineer
San Francisco, CA, USFull-time3+ years$110K - $160K0.25% - 1.50% equityVisa: Will sponsor
What the role involves

As a founding engineer, you'll help shape the foundation of how intelligent agents are built, verified, and deployed — not just at Relari, but across the next generation of AI-native software. In this role, you will: Architect core systems for Nuvi — designing how natural language specifications are turned into structured, testable agent behavior. Build infrastructure that enables agents to integrate with real-world tools, APIs, and knowledge sources. Ensure agent reliability by developing systems for simulation, automated evaluation, and behavioral verification. Work directly with customers, helping them go from intent to working agents — and using that feedback to inform what we build next. Continuously explore new ways agents can reason, act, and improve — and bring those innovations into the product. You’re a good fit if: You’ve built and shipped technically complex systems — whether infrastructure, AI tools, or end-user products. You’re a Software Engineer, Machine Learning Engineer, or Research Engineer at a fast-growing software company or top research lab. You thrive in a fast-moving environment and view unstructured environments as opportunities to identify the most impactful work and define the future success of the company. You operate like a owner: you identify high-leverage problems, drive them to resolution, and raise the bar for the team. Bonus points if you have: Built or contributed to an AI agent framework, orchestration layer, or evaluation pipeline. Trained or fine-tuned foundation models, or worked on tools that enable others to do so. Published in top ML/NLP conferences (e.g. NeurIPS, ICML, ACL) or contributed to widely used open-source projects.

PythonReactMachine learning

Roles are as last read from the company’s own listings. Openings close without notice — check the date on the listing before you spend an evening on the application.

Check the company’s own careers page — linked at the top — before a job board. A role appears there first, sometimes weeks before it is syndicated anywhere else.

Questions and experiences

Nobody has asked anything about Relari yet. If you have interviewed here, what you know is worth more to the next person than anything on the rest of this page.

Reviewed before it appears. Do not include anything that identifies you or anyone else.

Company facts compiled from public sources and last refreshed 9 September 2026. Details change; treat the company’s own site as the authority.

jobo is a browser extension. Open this on a computer to install it.