Research Engineer

Role overview Improve the reliability and usefulness of AI agents used across research, engineering, and business work. This applied research role combines production engineering with experiments: you will build ways to evaluate agent performance on real tasks, investigate failures, and use evidence to improve agents and their supporting systems. Over time, the work can expand toward more open-ended research and shared knowledge systems for agents.

Responsibilities - Build evaluation frameworks and experiments that measure agent performance, reliability, regressions, and trade-offs such as cost. - Work with engineering, quantitative research, and other teams to turn feedback about agent behavior into testable questions and measurable outcomes. - Diagnose failures across models, tools, prompts, and agent harnesses, then implement improvements. - Ship integrations and internal tools, and contribute to production agents and their supporting infrastructure. - Help create reusable knowledge and skills that agents can draw on and improve over time. - Collaborate with infrastructure and security partners to support dependable operation in production.

Requirements - Hands-on experience building, evaluating, or improving LLM models or agents through work, research, open-source contributions, or substantial projects. - Strong Python and software engineering skills, including attention to testing, maintainability, and reliability. - Experience with applied research, machine learning, or data science: designing experiments, working with data, and interpreting results. - Understanding of agent systems beyond prompting, including tool use, harnesses, context management, and common failure modes. - Ability to break open-ended problems into testable questions and work independently across research and engineering. - Experience building tools or systems for other users and assessing their effectiveness in practice.

Nice to have Experience with evaluation pipelines, benchmarks, reinforcement learning environments, agent frameworks, applied ML research, or model training. Open-source contributions, postgraduate study in a relevant field, financial or quantitative research knowledge, and familiarity with AI security are also useful. Prior crypto or decentralized finance experience is not required.

Benefits and work setup Remote work is available to people based in the contiguous United States or Canada. Listed benefits include paid medical, dental, and vision premiums, a home-office stipend, monthly connectivity reimbursement, unlimited vacation, paid parental leave, fertility benefits, and potential incentive compensation.