Remote job
Founding Harness Engineer
Job details
About this role
Role overview Build the runtime and reliability foundations for an AI coworker used in everyday team work. You’ll shape how the agent gathers context, chooses whether and how to act, uses tools safely, and follows tasks through to completion.
Responsibilities - Design the agent runtime, including context gathering, response decisions, tool selection, approval steps, and task completion. - Develop and refine system prompts so the agent is useful and appropriately cautious. - Create evaluation and regression tests for real tasks, tool use, and failure scenarios, and use them to assess prompt changes. - Improve how the agent handles long-running work, interruptions, and retries. - Establish safe execution boundaries, including approval gates and separation of data between users or tenants. - Work with leadership and early users to shape the product.
Requirements - Typically at least four years of relevant experience, with ownership of broad technical work from prototype through production. - Experience shipping production LLM agents that use tools, beyond demos or thin prompt wrappers. - Practical system prompt design experience, with testing and iteration. - Experience building evaluation and regression harnesses for LLM or agent behavior. - Understanding of safe tool execution, approval controls, and multi-user data separation. - Sound judgment about when an agent should act, request approval, or remain quiet, and ability to explain technical tradeoffs.
Benefits and work setup - Founding equity is available; the percentage and vesting terms are not yet specified. - Remote work from Serbia, Poland, or another Eastern European country. Working-hour overlap and employment or contractor arrangements are to be determined.