Senior AI Robot Engineer – Whole-Body Manipulation

I. About the Role

VinDynamics is developing next-generation humanoid robots capable of performing complex tasks in home and office environments.

We are seeking talented AI Robot Engineers specializing in Whole-Body Manipulation to join our team to develop and deploy AI models that enable humanoid robots to coordinate locomotion, balance, and manipulation to perform real-world tasks.

II. Key Responsibilities

• Whole-Body Manipulation: Develop learning-based policies that coordinate the hands, arms, and whole body for reaching, grasping, lifting, carrying, and other contact-rich manipulation tasks.
• Robot Learning: Apply reinforcement learning (RL), imitation learning (IL), vision-language-action (VLA) models, World Action Models (WAM), and motion tracking to develop robust humanoid manipulation and loco-manipulation skills.
• Simulation & Training: Build scalable training pipelines using Isaac Lab, Isaac Sim, MuJoCo, or similar simulators, including reward design, curriculum learning, and domain randomization.
• Human Motion Learning: Leverage human motion-capture data, teleoperation, demonstrations, and egocentric data for motion retargeting, imitation learning, and policy improvement.
• Sim-to-Real Deployment: Transfer trained policies to physical humanoid robots, optimize real-time inference, and improve robustness and task success rates in real-world environments.
• System Integration: Integrate AI models with whole-body controllers, motion planners, perception systems, and vision-language-action (VLA) models.

III. Qualifications

Minimum Qualifications

• Master’s or Ph.D. degree in Robotics, Computer Science, Artificial Intelligence, Control Engineering, or related fields.
• 2 years of experience applying multimodal data to train AI models for robotic manipulation.
• Strong understanding of data types and data requirements to train AI models for whole-body manipulation.
• Strong understanding of reinforcement learning, imitation learning, VLA models, diffusion-based methods, World Action Model (WAM),
• Solid understanding of robot kinematics and dynamics, and feedback control.
• Strong programming skills in Python, C , and PyTorch.
• Hands-on experience with robotic simulation platforms such as Isaac Lab, Isaac Sim, or MuJoCo.
• Experience developing and training policies for robot manipulation, locomotion, or whole-body control.
• Strong problem-solving skills and the ability to design, evaluate, and debug robotic learning systems.

Preferred Qualifications

• Experience deploying learning-based policies on physical humanoid, quadruped, or robotic manipulation platforms.
• Familiarity with PPO, actor-critic methods, imitation learning, motion tracking, and sim-to-real transfer.
• Experience with human motion retargeting, teleoperation data collection, and large-scale robot learning pipelines.
• Experience deploying AI models on real robots, including optimization for real-time inference.
• Experience with bimanual manipulation, dexterous hands, and contact-rich manipulation tasks.
• Familiarity with ROS 2, VLA models, diffusion policies, or robotics foundation models.
• Publications or open-source contributions in robotics, robot learning, or embodied AI are a plus.

IV. What You Will Build

• Whole-Body Reaching: Humanoid robots that bend, squat, step, and reach to manipulate objects.
• Loco-Manipulation: Robots that walk while grasping, lifting, and carrying objects.
• Bimanual Manipulation: Coordinated two-arm manipulation with balance-aware control.
• Real-World Autonomy: Robust whole-body skills integrated with perception and high-level task planning.