AI Harness Engineer (100% Remote, Worldwide)

Role overview This role focuses on building the next generation of agent infrastructure that lets local AI models run cooperatively across a peer-to-peer network. The work blends AI tooling, distributed systems, and agent harness engineering to make always-on, locally hosted AI assistants practical at scale.

Responsibilities - Ship new features and fixes for an agent harness layer, with a constant eye on reliability, latency, and user experience - Design and develop reusable modules and tools that unlock new AI capabilities across many model sizes and architectures - Implement and extend peer-to-peer technology so local AI workloads can coordinate without centralized servers - Maintain a high-quality codebase through coding standards, code review, and automated testing - Model complex state in JavaScript and TypeScript, from sync protocols to agent execution loops - Build evaluation and test infrastructure for non-deterministic AI behavior

Requirements - 5+ years of production software engineering, with deep experience in P2P systems, AI/LLM tooling, or both - Hands-on experience building agent harnesses, including tool calling, agent loops, context management, and structured outputs - Strong JavaScript and TypeScript skills, comfortable designing modular, maintainable code and publishing reusable packages - Experience designing for failure modes such as network partitions, model errors, timeouts, and retries - A self-described hacker mentality with a strong interest in P2P, distributed systems, or both - Comfort working remotely in a globally distributed team with clear written communication

Nice to have - Solid testing and debugging skills across complex async systems - Open-source contributions to P2P, networking, or AI tooling projects - Background in managing npm packages at scale