Research Fellow (Robot Learning & Manipulation)

About the role

The SHARE@NTU Corporate Laboratory, part of NTU, develops key technologies for humanoid robotics, covering multimodal sensing, artificial intelligence, environmental perception, and real-time motion planning. It is hiring a Research Fellow with a strong robotics learning background to develop and improve visuomotor manipulation policies, with a heavy emphasis on real-robot deployment.

Key responsibilities

• Lead research and development of learning-based visuomotor policies for humanoid robot manipulation, from problem formulation to real-world validation.
• Define research directions and technical roadmaps for grasping, object reorientation, bimanual manipulation and assembly.
• Advance behavior cloning, reinforcement learning and VLA-based reasoning, aiming for novel contributions and publishable outcomes.
• Address robustness challenges such as sensor noise, partial observability, contact dynamics and environment variability.
• Oversee the research pipeline from data collection and model training to evaluation and deployment.
• Manage milestones, deliverables and timelines, coordinate with collaborators and report progress to the PI.
• Mentor research associates, engineers and PhD students.
• Work with perception, controls, systems and hardware teams to integrate learned policies into the autonomy stack, and evaluate learning-based versus classical approaches.
• Lead preparation of publications, technical reports and grant/project documentation.

Required profile

• PhD in Robotics, Computer Science, Electrical Engineering or a related field.
• Strong publication record in robot learning, manipulation or embodied AI (e.g. CoRL, RSS, ICRA, IROS, NeurIPS, CVPR).
• Demonstrated experience developing and deploying robot learning systems on real robots.
• Experience supervising or mentoring junior researchers or students.

Required skills

• Deep expertise in robot manipulation and visuomotor control.
• Strong command of behavior cloning, reinforcement learning or related methods.
• Proficiency in Python and modern deep learning frameworks such as PyTorch.
• Ability to define research problems, design experiments and drive projects to completion.

Bonus qualifications: prior work on humanoids or dexterous platforms, experience moving prototypes to production systems, track record of industry collaboration, and passion for real-world humanoid robots. Only shortlisted candidates will be notified.