Position Summary:You will help build and refine the learned controllers that give our humanoid robots capable whole-body motion, and help get them working reliably on real hardware. Working alongside senior RL and controls engineers, you'll train and tune policies, run experiments in large-scale simulation, and take an active role in bringing controllers up on physical robots and closing the sim-to-real gap. This is a hands-on role for an engineer who wants to grow their depth in reinforcement learning and control while contributing directly to robots that work in the real world-implementing, tuning, evaluating, and debugging the controllers that others help architect, and taking increasing ownership over time.
Core Responsibilities:- Policy Training & Tuning: Train, tune, and evaluate reinforcement learning policies for whole-body control tasks such as balance, locomotion, and manipulation.
- Simulation Experiments: Set up and run experiments in large-scale simulation, analyze results, and iterate quickly on policy designs.
- Sim-to-Real Support: Contribute to the sim-to-real pipeline-domain randomization, system identification, and the iterative work of making policies transfer to hardware.
- Real-Robot Bring-Up: Help bring controllers up on physical robots, run experiments, and debug behavior in the real world.
- Evaluation & Tooling: Build and improve tools for evaluating controller performance in simulation and on hardware.
- Collaboration: Work closely with senior RL, controls, and hardware engineers, taking direction on harder design decisions while owning your own workstreams end-to-end.
Required Qualifications:- Candidates should have strong foundations in at least one of the following
- RL: Solid working knowledge of reinforcement learning - including policy optimization methods (e.g., PPO), reward shaping, domain randomization, demonstration-guided RL - with hands-on experience training policies for continuous control tasks such as locomotion or manipulation.
- Model-Based Control: Solid working knowledge of optimization-based control, such as trajectory optimization, MPC, QP-based control, inverse kinematics and differential IK.
- Foundational understanding of robot modeling: dynamics, kinematics, coordinate frames, and control fundamentals (for instance PID control, stiffness/impedance control).
- Real-Robot Experience: Experience working with physical robots-bringing up, testing, or debugging controllers on real hardware.
- Software Engineering: Strong Python skills and the ability to write clean, testable code; working C++ ability or willingness to grow it.
- Simulation: Experience with setting up, implementing and, and running experiments in physics simulation for robotics or RL, for instance in Drake, Mujoco, etc.
- Curiosity & Drive: Eagerness to learn, iterate quickly, and take on increasing ownership.
Preferred Qualifications:- Experience with whole-body control, locomotion, or manipulation on legged or humanoid robots.
- Familiarity with sim-to-real techniques and their practical challenges.
- Exposure to classical/model-based control (MPC, QP controllers, trajectory optimization).
- Experience with ML experiment-tracking and training workflows.
- Experience with state-of-the-art machine learning and GPU programming frameworks, such as pytorch, WARP, JAX.
- Coursework, projects, or publications in robotics, RL, or control.
Walden Robotics offers a competitive total compensation program, including salary, annual cash bonus, company equity, company-subsidized insurance programs, 401(k) with company match, flexible PTO, daily lunch, and other benefits. The pay ranges noted on our posts are for salary only.
The pay range for this role is:
157,500 - 210,000 USD per year (BOS)
