Principal AI Research Engineer - RL
Reflex Roboticsover 1 year ago
San Francisco, CA, USA or New York, NY, USAStaff+
Responsibilities
- Develop robust on-policy reinforcement-learning policies for humanoid robots.
- Re-implement and debug core reinforcement-learning algorithms, including SAC and DDPG.
- Tune hyperparameters and address unstable gradients during training.
- Contribute intellectual advances to sample-efficient reinforcement-learning algorithms such as DreamerV3 and MuZero.
- Deploy and ship policies that learn and operate on real robot hardware.
Requirements
- Experience re-implementing core reinforcement-learning algorithms such as SAC and DDPG from scratch.
- Ability to debug unstable gradients and tune reinforcement-learning hyperparameters correctly.
- Meaningful intellectual contributions to sample-efficient reinforcement-learning algorithms such as DreamerV3 and MuZero.
- Experience shipping on-policy reinforcement learning on physical hardware, such as quadruped walking or drone racing.
- A portfolio or substantial technical project work is welcomed.
Benefits
- High ownership and high equity at a small robotics startup.
- Opportunity to shape a humanoid robotics product and autonomy stack from the ground up.
- Work on physical robots deployed for real industrial customer problems.
Categories
AI ResearchRobotics
About Reflex Robotics
Reflex Robotics builds low-cost wheeled humanoid robots to automate dangerous and repetitive work in manufacturing and logistics. The company designs and manufactures its systems in-house, coupling hardware with AI trained from large-scale demonstration data, and sells robots to industrial operators. Founded in 2022 and headquartered in Brooklyn, New York, it is privately held and raised seed funding in 2023; its core team previously led work on Boston Dynamics’ Stretch and Tesla production lines.