AI Harness Engineer (100% Remote, Worldwide)

Role overview Build the software systems that connect people to capable local AI assistants. You’ll develop agent tooling and peer-to-peer features, helping AI run across different models and devices while keeping the product reliable and responsive.

Responsibilities - Deliver product features and fixes with attention to reliability, performance, and user experience. - Design tools and modules that support AI capabilities across models of different types and sizes. - Build and improve peer-to-peer technology for local AI use. - Maintain code quality through shared standards, reviews, and automated tests. - Account for failures such as network interruptions, model errors, timeouts, and retry behavior.

Requirements - At least five years of experience building production software. - Strong JavaScript or TypeScript skills, including modelling complex state in systems such as sync protocols or agent loops. - Practical experience building agent harnesses, including tool calls, agent loops, context management, and structured outputs. - Experience creating evaluations and test infrastructure for systems with non-deterministic behavior. - Deep experience in peer-to-peer systems, AI or LLM tooling, or both. - Experience designing software to handle failures gracefully.

Nice to have - Open-source contributions involving peer-to-peer systems, networking, or AI tooling.

Benefits and work setup The role is described as fully remote and open worldwide.