Physical AI Research Scientist Intern

Responsibilities

• Research and develop Physical AI including VLA, World Models, and Robot Foundation Models
• Analyze latest papers and derive improvements for internal model architecture, training, and evaluation
• Collect, refine, and analyze real, simulation, and web-based robot data
• Build evaluation frameworks in simulation and real robot environments
• Perform Sim-to-Real validation and analyze real-world performance degradation
• Design in-house benchmarks and task suites reflecting manufacturing conditions
• Improve generalization across environments, tasks, objects, platforms, and language instructions
• Develop closed-loop data collection and re-collection strategies from real robot evaluations

Requirements

• Master's or PhD candidate in Physical AI, Robotics, Computer Vision, ML, Multimodal AI, or Embodied AI
• Research experience in robot manipulation, imitation learning, RL, VLA, World Models, or Embodied AI
• Experience with large-scale model training, fine-tuning, data mixing, filtering, or evaluation automation
• Hands-on experience evaluating models in real robot or simulation environments

About the Company

LG AI Research's Physical Intelligence Lab develops visual intelligence technologies that surpass human vision, creating new experiences in real and virtual spaces. The lab researches computer vision, multimodal learning, 3D understanding, and robotics to build core vision technologies for the LG Group and lead future industries.