Mirendil

Mirendil

Member of Technical Staff, Agent Harness

San Francisco · Staff+

Sponsorship not specified$300k-$500kDetected 28 days ago
Research

About the role

  • We are looking for an engineer who is passionate about giving the model tools to perform the best it can.
  • If you have a penchant for building your own tools, this role will be a good fit.
  • Some example areas you might work on:

Responsibilities

  • We want people who deeply understand model capabilities and can build efficient architectures for the model and researchers to work across.
  • Build and innovate on the agent harness: agent loop architecture, tool integrations, prompt scaffolding, execution environments, and capability primitives
  • Design orchestration systems for horizontal scaling of agents: memory, state management, multi-agent coordination, and task decomposition
  • Build guardrails and reliability mechanisms that make long-horizon agentic tasks robust across failures, unexpected model behavior, and edge cases
  • Develop evaluation and observability tooling so the team can measure agent behavior, catch regressions, and iterate quickly
  • Own the extension layer between our models and external tools, APIs, and environments - making it fast to bring new capabilities online

Compensation

  • We offer a base salary of $300,000–$500,000 USD and a meaningful equity grant, depending on experience and background, along with competitive benefits.

This listing is sourced directly from Mirendil's careers page and normalized into a canonical job model.