1G06. [Contract] Embodied AI Model Development Research Engineer (R&D Business)English summary from Japanese

Must-haves

• Practical experience in Machine Learning / Deep Learning, including developing models and building training pipelines with frameworks like PyTorch.
• Software development experience using Python or C++.
• Strong interest in robotics and applying AI models to the physical world, with a willingness to learn basic concepts like coordinate transformations.
• Strong motivation for social implementation, prioritizing practical usability over theoretical score improvements.

Nice-to-haves

• Research and development experience with World Models or video generation models, or knowledge/experience in predicting/planning environment dynamics using them.
• First-author publication record in top-tier conferences (NeurIPS, ICLR, CVPR, ICML, CoRL, ICRA, IROS, etc.).
• Experience with middleware like ROS/ROS2, or robot control/API integration experience using Python.
• Experience handling large-scale robot operation datasets (e.g., Open X-Embodiment), open VLA models (e.g., π0, OpenVLA), or frameworks (e.g., openpi, lerobot).
• Experience building reinforcement learning/imitation learning environments using physical simulators (e.g., NVIDIA Omniverse/Isaac Sim, MuJoCo, Habitat).
• Experience developing and socially implementing AI models like LLM/VLM for specific domains (e.g., manufacturing, healthcare, finance).

Tech stack

Work style

Hybrid. Work from home or the office in Tokyo. May require 1-3 days of in-office work per week depending on the role's tasks.

Other notes

Hourly wage: 5,000 JPY - 8,000 JPY. Determined based on experience and skills. Contract-based position.