Applied Research - Robotics

Prime Intellect•San Francisco, CA

1d•$150,000 - $300,000

About The Position

Prime Intellect is hiring an applied researcher to work with an external robotics team on environments, evaluations, agent systems, and post-training methods. The role involves training and evaluating agents using reinforcement learning, world modeling, imitation learning, planning, and multimodal perception, with feedback from simulation and deployed robots. Experiments will examine behavior under partial observations, long horizons, changing environments, and hardware and safety constraints. Applied researchers turn workflows into tasks, environments, evaluations, and reward signals; run post-training experiments against real baselines; and determine whether observed gains came from the model, data, harness, inference, or infrastructure. Reusable work can become shared research, open-source systems, and product capabilities. The position is externally facing, and researchers typically work directly with customers, users, or research partners while maintaining a substantive research agenda.

Requirements

Relevant experience in robotics, embodied AI, reinforcement learning, imitation learning, control, world models, simulation, computer vision, agent evaluation, or production systems interacting with the physical world.
Hands-on experience building and debugging model-training, post-training, agent, evaluation, or closely adjacent experimental loops.
Strong experimental judgment, including meaningful baselines, careful trace inspection, clear failure attribution, and willingness to change direction when evidence changes.
Strong Python skills and ability to build artifacts that other researchers and engineers can run, inspect, and extend.
Comfort owning ambiguous problems and keeping partners and teammates current while resolving them.

Nice To Haves

Evidence of exceptional work, such as publications, open-source systems, benchmarks, or shipped products.
Ability to show a difficult project built, what failed, and how evidence changed the approach.

Responsibilities

Build simulation and real-world environments for training and evaluation, and conduct careful sim-to-real comparisons.
Develop long-horizon agents that perceive, plan, act, remember, and recover under hardware constraints.
Develop training approaches combining offline and online reinforcement learning, demonstrations, interventions, and outcome feedback.
Create rewards and evaluations for task success, safety, adaptation, and continual improvement without silent regressions.
Build aligned simulation and real-world evaluations for long-horizon embodied tasks and measure where policies fail to transfer.
Train embodied agents to recover from partial failures, changing scenes, and uncertain state while respecting safety and hardware constraints.
Run post-training experiments against meaningful baselines and investigate whether gains came from the model, data, harness, inference, or infrastructure.
Work directly with customers, users, or research partners while maintaining a substantive research agenda.