AI Research Scientist, Video Understanding & Physical AI

AI Research Scientist, Robotics & Embodied AI

Samsung SDS America AI Science Lab is researching the next generation of intelligent robotic systems that perceive, reason, learn, and act in the physical world. Working with multiple robot hardware partners, we develop complete embodied AI systems spanning data collection through teleoperation, learning and policy development, simulation, and deployment on physical robots in real-world environments.

We are looking for an AI Research Scientist, Robotics & Embodied AI to join our team and help advance the science behind intelligent robotic systems. You will research and develop learning-based approaches that enable robots to understand their environment, learn complex behaviors, manipulate objects, and perform tasks autonomously.

This role sits at the intersection of artificial intelligence, machine learning, robotics, computer vision, and embodied intelligence. You will work on problems such as robot learning, imitation learning, reinforcement learning, vision-language-action models, multimodal perception, dexterous manipulation, and learning-based control.

Unlike a traditional research role focused primarily on simulation or publications, you will have the opportunity to take research from hypothesis to experiment to physical robot deployment. You will work directly with humanoid and other robotic platforms, use teleoperation and real-world demonstrations to generate training data, develop and evaluate learning-based policies, and iterate through experiments on physical hardware.

What sets our lab apart is the combination of research freedom and real-world impact. You will work alongside a small, highly technical team where researchers have direct ownership of their work—from defining research questions and developing novel approaches to implementing experiments and evaluating results on real robots.

Samsung SDS builds and operates systems supporting manufacturing at scale across thousands of factory lines worldwide. The embodied AI capabilities we research today are being developed with this production environment in mind, with a long-term vision extending from individual intelligent robots to coordinated multi-robot systems operating in complex industrial environments.

If you are excited about advancing AI from systems that understand the digital world to systems that can perceive, reason, learn, and act in the physical world, this is an opportunity to do research that moves directly into reality.

This position is full time onsite in Mountain View, CA. This is a must to be considered.

Responsibilities

• Conduct original research in robotics, embodied AI, physical AI, and machine learning
• Develop novel algorithms and learning approaches for intelligent robotic systems
• Research and develop methods for robot learning, imitation learning, reinforcement learning, and learning-based control
• Explore vision-language-action and multimodal models for robotic perception, reasoning, and action
• Develop approaches that enable robots to learn complex manipulation, grasping, locomotion, and interaction behaviors
• Design experiments using real-world robot platforms, simulation environments, and large-scale robotics datasets
• Develop and evaluate data collection pipelines using teleoperation, demonstrations, and other forms of human-guided robot interaction
• Train, evaluate, and deploy learning-based policies on physical robotic platforms
• Research dexterous manipulation, grasping, and contact-rich interaction for humanoid and other robotic systems
• Investigate methods for transferring learned behaviors from simulation and synthetic data to real-world robots
• Develop scalable training and evaluation pipelines for embodied AI models
• Integrate research algorithms with robot perception, sensors, actuators, and low-latency control systems
• Analyze experimental results, identify limitations, and formulate new research directions
• Collaborate with robotics engineers, AI researchers, and hardware teams to translate research concepts into working robotic systems
• Contribute to research publications, technical reports, patents, and intellectual property where appropriate
• Stay current with advances in embodied AI, foundation models, robot learning, computer vision, reinforcement learning, and related research areas
• Contribute to the architecture and long-term research direction of next-generation embodied AI systems Basic Qualifications