Senior Engineer, Agent Runtime

Role overview Own the local runtime that carries out an agent’s tool calls, supervises processes, and records sessions. The work centers on defining clear limits between an agent’s requests and actions performed on a user’s machine, while keeping runtime behavior consistent across major desktop operating systems.

Responsibilities - Design and ship runtime policy layers, including a possible capability model for MCP servers, and document their limits. - Make behavior consistent across Windows, macOS, and Linux, with CI coverage on all three platforms. - Design an append-only session log that supports rewinds, checkpoints, and exports without unnecessary cost. - Review the work of other runtime engineers and take part in peer review. - Work directly in the runtime code when documentation does not answer implementation questions.

Requirements - Extensive practical TypeScript experience, ideally with Bun or Node.js. - A track record of shipping software that other engineers run locally, such as a CLI, build tool, or agent. - Ability to reason carefully about the boundary between requested actions and what a machine actually executes. - Strong written English for design notes that communicate decisions clearly. - Comfort working across Windows, macOS, and Linux and proving behavior through automated CI.

Benefits and work setup - Remote role based in Europe. - The hiring process includes a short introduction and work samples, a team conversation, a paid take-home exercise expected to take an afternoon, and a final conversation. The process is intended to take about two weeks, with replies at each stage.