AI Evaluation Engineer

About the company Our client is a software company. The role Raydar is recruiting for this role on behalf of our client. Lead the creation of a measurement approach for an AI system that carries out technical design work. Draw on both hands-on engineering judgment and evaluation know-how, and collaborate with domain experts and other teams to convert findings into clear improvement priorities. What you'll do - Establish a new assessment framework that defines quality for design work where several answers can be valid. - Collaborate with domain experts, external users and internal teams to see how seasoned engineers operate and where the system struggles. - Reproduce realistic design tasks and typical failure cases in repeatable test setups that developers can run again. - Create and tune automated graders alongside human review processes, and confirm that the graders agree with expert opinion. - Convert assessment results into a ranked list of improvements to the system's reasoning. Requirements What we're looking for - About five or more years of professional experience in hardware engineering, AI, or a blend of both, with depth in one and practical familiarity with the other. - Substantial hands-on background assessing AI systems, including authoring test tasks, building automated graders and benchmarking them against human review. - Solid grasp of how autonomous, agent-style AI systems behave. - Grounding in a hardware engineering field, enough to tell sound design from weak design. - Comfort working directly with external users, specialists, partners and technical colleagues across functions. - Self-assurance to guide work among many stakeholders in a role with high ownership. - Based in the US or Canada, and able to travel to an office for one week every 6 weeks. Bonus points - Prior time at a robotics or physical AI company. - Experience building things from zero in an early-stage startup. Benefits Compensation and benefits - Base salary: USD 200,000 to 300,000 per year - Equity Location and work model - Remote (US or Canada) - Full-time Engineering Computer Software Mid-Senior level Full-time Remote role listed by employer.