Senior Researcher, Robot Intelligence

About the role

The will own the full lifecycle of robotic intelligence, transforming high-level concepts into robust systems that operate reliably in diverse, real-world settings. You will define technical direction, execute current research, and partner across academia and industry to push the boundaries of general-purpose robotics. Your contributions will directly shape Samsung's long-term presence in robotics and create meaningful user impact by ensuring that intelligent machines perceive and act with reliability in everyday human environments. You will architect and validate multimodal sensing and control systems tailored for next-generation home robots, focusing on making embodiment dependable at scale. A core focus will be leveraging vision language models to enable dexterous manipulation and adaptive behavior in unstructured and semi-structured environments where prior assumptions often fail. You will design and execute large-scale simulation campaigns and real-world experiments to refine embodied intelligence, closing the loop between open-world 3D perception, decision-making, and task execution. Collaboration is integral to this role, and you will translate insights from reinforcement learning into reliable behaviors that function outside the lab. A key part of your impact will come from producing high-quality scientific publications and patents that strengthen Samsung's thought leadership in robotics and AI.

Key facts

What you'll do

• Architect and validate multimodal sensing and control systems tailored for next-generation home robots, ensuring robustness across diverse household conditions.
• Leverage vision language models to enable dexterous manipulation and adaptive behavior in unstructured environments where objects, layouts, and users vary widely.
• Design and execute large-scale simulation campaigns in platforms such as Mujoco and Isaac Sim to stress-test policies and refine embodied intelligence before real-world deployment.
• Conduct real-world experiments that close the loop between open-world 3D perception, decision-making, and task execution to validate performance at scale.
• Collaborate with universities, startups, and government research labs to align laboratory innovations with real-world product timelines and feasibility.
• Translate insights from reinforcement learning into reliable behaviors that function in semi-structured and unstructured settings, handling ambiguity and partial observability.
• Produce high-quality scientific publications and patents that strengthen Samsung's thought leadership and establish technical credibility within the global robotics community.
• Partner with product teams to embed foundation models into Samsung's consumer electronics portfolio, ensuring that research outcomes translate into user-facing value.
• Ensure data quality and robust model monitoring throughout the development lifecycle to maintain safety, performance, and compliance with operational standards.
• Define and own the technical roadmap for robot intelligence, balancing long-term research ambitions with near-term product constraints and opportunities.
• Evaluate emerging techniques in foundation model training, inference, and alignment to identify opportunities for competitive advantage in embodied systems.
• Mentor and collaborate with cross-functional engineers and researchers to elevate the overall technical maturity of the robotics organization.
• Implement rigorous methods to monitor model performance and data quality across training, validation, and deployment stages.
• Optimize system-level efficiency by co-designing algorithms, sensor suites, and compute resources to meet real-world power and latency constraints.
• Act as a technical authority within Samsung and in external partnerships, representing the Robot Intelligence division in key strategic discussions.

Requirements

• Hold a PhD in Robotics, Computer Vision, or a closely related field, with a demonstrated focus on embodied intelligence and real-world robotic systems.
• Bring a minimum of two years of industry experience in robotics foundation models, vision language action systems, and vision language models applied to physical environments.
• Demonstrate a strong grasp of computer vision, including object detection, segmentation, tracking, and multi-view image processing across diverse lighting and motion conditions.
• Possess hands-on experience with reinforcement learning, imitation learning, task planning, and simulation platforms such as Mujoco and Isaac Sim for scalable policy development.
• Apply expertise in foundation model techniques including contrastive learning, fine-tuning, and LoRA within generative AI and diffusion models for perception and control.
• Use rigorous methods to monitor model performance and data quality throughout the entire development lifecycle, from data collection to deployment.
• Demonstrate technical fluency with PyTorch, TensorFlow, C, C++, Java, ROS, and CUDA to build performant and maintainable robotic software stacks.
• Possess deep knowledge of robotics fundamentals such as locomotion, manipulation, sensor inputs, pose tracking, 3D mapping, SLAM, and policy architectures designed for semi-structured and unstructured environments.
• Operate effectively both independently and within cross-functional teams, communicating complex results with clarity and precision to both technical and non-technical stakeholders.

Nice to have

No specific "nice to have" skills are listed beyond the core qualifications. All tools and technologies relevant to the role are described in the responsibilities and qualifications sections.

Practical notes

Please confirm all details on the official application page before submitting your materials.