Open Role
Embodied AI Algorithm Engineer (Robot Manipulation)
Singapore
Develop and deploy end-to-end manipulation models for humanoid and mobile platforms.
Responsibilities
- Develop end-to-end robot manipulation models for humanoid robots and mobile manipulation platforms.
- Design, train, fine-tune, and deploy Vision-Language-Action (VLA), World Action Models (WAM), and related embodied AI systems.
- Explore methods for transferring human demonstration data (e.g., UMI systems) to whole-body robotic manipulation.
- Analyze and curate large-scale datasets to improve model performance, robustness, and generalization.
- Work closely with robotics, data, and infrastructure teams to deploy models in real-world environments.
Qualifications
- A deep and genuine passion for Embodied AI is essential.
- Master's degree or above in Computer Science, Robotics, Automation, or related fields.
- Strong experience with deep learning frameworks such as PyTorch, TensorFlow, or JAX.
- Hands-on experience training and deploying robot manipulation models such as Diffusion Policy, OpenVLA, π0, or similar architectures.
- Familiarity with World Models, Reinforcement Learning, and multimodal foundation models.
- Experience with robotic data collection systems and real-world robot deployment is highly preferred.
- Experience with UMI-based data collection and manipulation learning is a plus.