Research Engineer, Hardware Intelligence

Nominal•Los Angeles, CA

12h•Onsite

About The Position

Nominal is building a Hardware Intelligence Platform to help engineers gain visibility into complex physical systems, make better decisions, and bring products into the real world faster. Its platform supports hardware engineering teams in simulating, testing, manufacturing, and operating systems across sectors including aerospace, defense, manufacturing, transportation, energy, and robotics. The Hardware Intelligence team is developing agents expert in engineering physical systems. The role will set and execute the research agenda, using real hardware test results to assess agent performance and improve agent context, tools, and models. The engineer will investigate failures, build evaluations and training resources grounded in measured results, develop research infrastructure, test improvement approaches, and deliver successful work into customer-facing agents.

Requirements

At least 3 years of experience in ML engineering or research; less experience may be acceptable for candidates whose recent work directly matches this role at a frontier lab or AI-native company.
Hands-on experience improving an LLM or agent system through evaluations, data, post-training, or harness changes, and demonstrating that the improvement worked.
Strong engineering ability, including writing pipelines, training scripts, and graders.
Experimental judgment, including designing sound experiments and determining whether observed differences are real.
Comfort choosing the next experiment when the outcome is uncertain.
Daily use of modern AI coding agents such as Claude Code, Cursor, or Codex, and curiosity and openness to changing tools and work practices.

Nice To Haves

Experience with post-training on real tasks, including supervised fine-tuning, reinforcement learning from feedback or verifiable rewards, or distillation.
Experience building reinforcement-learning environments, synthetic-data pipelines, or benchmarks used by others.
Experience with computer-use agents, multi-agent systems, simulation, or edge deployment.
Experience with time-series, sensor, or scientific data, or models built over such data.
Experience as an engineer in hardware, testing, controls, or robotics, or close collaboration with those teams.

Responsibilities

Identify agent failures on real hardware work using customer traces and tasks designed with hardware experts.
Convert failures into evaluations, environments, and training data grounded in real test results.
Build graders and verifiers that check agent work against measured data, and identify where those graders and verifiers fail.
Improve agent performance through changes to prompts, tools, or context; fine-tuning; reinforcement learning; or distillation of open-weight models.
Prototype research approaches involving hardware context for agents, computer use in engineering tools, links between simulation and physical tests, collections of coordinated agents, and intelligence at the edge.
Build research infrastructure, including evaluation harnesses, dataset and trace pipelines, and training and experiment tooling.
Distinguish meaningful gains from noise, deploy successful improvements into customer-facing agents, and document findings.