Menlo Research
Researcher, Manipulation
Singapore +1 · Hybrid · Full-time · Posted Sep 4, 2026
- Location
- Singapore
Palo Alto
About the job
Develops and deploys reinforcement-learning manipulation skills for the Asimov humanoid robot, transferring them from simulation to real hardware.
What you'll do
- Design, train, and deploy policies for manipulation, grasping, dexterous tasks, and whole-body coordination.
- Own the manipulation sim2real pipeline, including MuJoCo environments, rewards, and randomization strategies.
- Use real-robot telemetry and failures to improve policies; collaborate across locomotion, controls, and hardware research.
What they ask for
- Hands-on experience with reinforcement learning or learning-based methods for manipulation, grasping, or dexterous control.
- Track record of deploying manipulation policies on real robots.
- Strong command of a physics simulator such as MuJoCo or Isaac, including contact modeling, reward shaping, and domain randomization.
- Fluency in Python and modern RL tooling; comfort with a ROS2-based control stack.
Nice to have
- Published work in manipulation, dexterous control, grasping, or imitation learning.
- Experience with tactile sensing, force control, teleoperation, whole-body control, or open-source robotics/RL projects.
Tools
MuJoCo, Isaac, Python, ROS2
AI focus
The role develops reinforcement-learning policies for robotic manipulation and uses AI as a collaborator in research and work.
Model-written summary · the original posting is authoritative