Researcher, Manipulation

About This Opportunity

We are representing a confidential robotics research organization that is building at the cutting edge of humanoid manipulation and reinforcement learning. This team works across the full stack—from simulation to real hardware deployment—and prioritizes shipping working systems over theoretical perfection. If you want to see your work move from code to physical manipulation tasks on real robots, our client offers that rare opportunity.

We are seeking a Researcher, Manipulation to design and deploy RL policies for grasping, dexterous tasks, and whole-body coordination. You will own the sim-to-real pipeline for manipulation, working in close collaboration with locomotion and hardware teams. This role is based in Singapore or the United States.

Responsibilities

• Design, train, and deploy reinforcement learning policies for manipulation, grasping, and contact-rich dexterous tasks on humanoid hardware
• Extend existing locomotion-to-manipulation RL frameworks toward whole-body manipulation where balance and reaching are integrated challenges
• Own the sim2real pipeline for manipulation using zero-shot transfer approaches so skills move cleanly to real hardware
• Design reward functions, training environments, and domain randomization strategies in physics simulators for contact-heavy tasks
• Close the loop with real telemetry from deployed systems, feeding failure modes back into policy iteration
• Collaborate with locomotion, controls, and hardware researchers on integrated humanoid behaviors
• Contribute to open-source projects and share learnings with the broader robotics community

What We're Looking For

Required:

• Deep hands-on experience with reinforcement learning or learning-based methods for manipulation, grasping, or dexterous control
• Proven track record of deploying manipulation policies onto real robotic systems, not just simulation or research papers
• Strong command of physics simulators such as MuJoCo, Isaac, or similar, including contact modeling, reward shaping, and domain randomization
• Fluency in Python and modern RL tooling; comfort working within ROS2-based control stacks
• Bias for shipping: preference for iterating on real hardware over perfecting in simulation
• Clear analytical thinking about contact-rich task failures and practical approaches to closing sim2real gaps

Nice to Have:

• Published work in manipulation, dexterous control, grasping, or imitation learning
• Experience with tactile sensing, force control, teleoperation, or demonstration data
• Experience with whole-body control spanning locomotion and manipulation
• Contributions to open-source robotics or RL projects

How We Work With You

Candidates apply directly through Get A Job.ai. Our talent team will conduct an initial screening to understand your background and ensure alignment with the role requirements. Qualified candidates are then submitted to our client for their review and interview process. Please do not attempt to contact the client directly, as all communication is managed through our recruiting team to ensure a smooth and professional process.

Pay

Compensation details will be discussed during the screening process and are competitive with senior research roles in robotics and applied machine learning.

Equal Opportunity: Get A Job.ai partners with employers who value diverse perspectives and do not discriminate on the basis of any legally protected characteristic. We encourage applications from candidates of all backgrounds. If you need accommodations during the application process, please inform your Get A Job.ai recruiter.