The Role:
Raydar is recruiting for this role on behalf of our client. Serve as the first hire in a new robotics research function, running experiments that train and evaluate robot control models in simulation. You will convert experimental findings into clear requirements for engineering colleagues, share results externally, and later build out a local team.
What You'll Do:
Train robot control policies using fine-tuning, imitation learning and reinforcement learning in simulation, starting with manipulation and grasping tasks.
Set baselines on public benchmarks, then retrain under identical protocols with added synthetic data and report what improved and what did not.
Turn experiment outcomes into written requirements for the data engineering group, including the kinds of scenes, the level of realism and the range of variability needed.
Present results at robotics conferences and demos, and join technical conversations with external partners.
Begin as an individual contributor and hire and lead a regional robotics team as it grows.
What We're Looking For:
MS or PhD in robotics, machine learning or a related field.
Hands-on experience training and evaluating robot policies in simulation environments.
Experience using evaluation results to decide data or simulation changes for the next training round.
Background in manipulation or grasping policy training.
Familiarity with policy learning methods such as vision-language-action fine-tuning, imitation learning or reinforcement learning in simulation.
Experience running controlled comparisons with fixed seeds, checkpoints and benchmarks while varying only the training data.
Self-directed approach to designing experiments and owning work end to end.
Bonus Points:
Graduate degree from a leading robotics program.
Experience with navigation policy training in simulation.
Location and Work ModelSan Francisco, CA, United StatesHybrid, 3 days per week in office
Full-time
Compensation, As The Posting Describes ItEquity
