Job Description

About Dexmate

Dexmate is an early-stage robotics startup building multi-purpose mobile robots capable of complex manipulation tasks. The goal is to advance robot dexterity to the point where a single platform can handle a diverse, general set of physical skills in real-world environments.

The founding team is small, research-driven, and looking for people who want to work at the frontier of robot learning — not just publish on it. This is an on-site role in Santa Clara where you will work directly with researchers and engineers building systems that run on real hardware.

The Opportunity

Dexmate is looking for a Research Scientist to join the founding team and help advance the core capabilities of their manipulation platform. You will conduct research across robotics, reinforcement learning, imitation learning, perception, and foundation models, and implement what you build on real robots.

This is not a role where research stays in simulation. You will work with real hardware, real failure modes, and a team that moves from idea to experiment quickly. If you want your research to shape the direction of a company and run on physical systems, this is that role.

What You'll Do

• •Develop new algorithms and methods for training AI models that enhance robot dexterity
• •Conduct research across multiple disciplines including RL, imitation learning, control, perception, and foundation models (LLM, VLM)
• •Work with large-scale ML systems and large-scale model training and fine-tuning
• •Design and implement state-of-the-art learning-based manipulation, navigation, and control algorithms on real robots
• •Collaborate with engineering and research teams to develop a diverse set of robust manipulation skills

You Should Have

• •Ph.D. in Robotics, Computer Science, Electrical Engineering, Mechanical Engineering, or a related field, or equivalent research experience
• •At least 3 years of experience conducting independent research
• •Deep understanding of state-of-the-art robot learning techniques including reinforcement learning and imitation learning
• •A track record of research excellence with publications at top venues such as RSS, CoRL, ICRA, NeurIPS, ICML, ICLR, CVPR, Science Robotics, or IJRR
• •Proficiency in Python and deep learning frameworks such as PyTorch, TensorFlow, or JAX
• •Hands-on experience with real robot experiments
• •Experience with robot simulators such as Isaac Gym, Isaac Sim, MuJoCo, SAPIEN, or Drake

Nice to Have

• •Experience building manipulation or navigation systems that generalize across tasks or environments
• •Interest in and experience with foundation models applied to physical robot control