AI Researcher (Embodied Robotics)

Sony Electronics Singapore; Singapore; On-site; SGD 90,000 - 150,000; Full time

Job summary

Pentas Vision seeks an AI Researcher to explore sensor applications in embodied robotics, turning academic innovations into real-world robotic products. You will advance AI systems that fuse vision, sensors, and control to enable autonomous robots in embedded devices.

Required expertise includes deep learning for embodied AI, multimodal perception, and hands‑on experience with simulators (Omniverse, MuJoCo, Isaac Sim, PyBullet).

Qualifications

• Master’s or Ph.D. in Computer Science, Robotics, Electrical/Mechanical Engineering, or a related field with focus on AI, robotics, or embodied intelligence.
• Familiarity with deep learning architectures for embodied AI, with experience in reinforcement learning, imitation learning, or multimodal perception.
• Experience in integrating hardware sensors into robotic systems, with ability to debug both software and hardware pipelines.
• Hands‑on experience with simulation platforms (Omniverse, MuJoCo, Isaac Sim, PyBullet).
• Proficiency in Python and C/C++, with experience in ROS, PyTorch, or TensorFlow.
• Strong problem‑solving skills, curiosity, self‑motivation, and passion for tackling challenging problems in robotics.
• Strong teamwork and communication skills.

Responsibilities

• Conduct frontier research in large-scale embodied AI models, including data generation, training optimization, and evaluation.
• Track and assess state-of-the-art research, and maintain/improve algorithmic frameworks.
• Design and implement proof-of-concept systems, validating them in both simulation platforms and real robotic hardware.
• Develop and deploy algorithms that leverage multimodal sensor data to enhance robotic intelligence and autonomy.
• Collaborate with cross-functional teams to integrate algorithms into robotic platforms and ensure robust real-world performance.
• Work closely with hardware and sensor teams to understand sensor characteristics and translate them into algorithmic advantages.
• Contribute to system integration and algorithm innovation, validating new sensors through cutting-edge AI benchmarks.

Skills

Deep learning for embodied AI

Reinforcement learning

Imitation learning

Multimodal perception

Python

C/C++

ROS

PyTorch

TensorFlow

Hardware‑software integration

Teamwork

Education

Master’s degree in Computer Science/Robotics/EE/ME

Ph.D. in Computer Science/Robotics/EE/ME

Tools

Omniverse

MuJoCo

Isaac Sim

PyBullet

ROS

CUDA

PyTorch

TensorFlow

Job description

On behalf of Pentas Vision, a subsidiary company of Sony Semiconductor Solutions Corporation (SSS), we are seeking for a AI Researcher to explore the applications of sensors in the field of embodied robotics, transforming academic innovations into real-world robotic products.

Pentas Vision specialises in the research and development of AI and in particular – Computer Vision technologies. Our core expertise lies in integrating AI solutions with diverse Image Sensors, facilitating seamless incorporation into embedded devices like mobile phones, IoT cameras, and other relevant devices.

What you will be doing:

• Conduct frontier research in large-scale embodied AI models, including data generation, training optimization, and evaluation.
• Track and assess state-of-the-art research, and maintain/improve algorithmic frameworks.
• Design and implement proof-of-concept systems, validating them in both simulation platforms (e.g., MuJoCo, Isaac Sim, Omniverse, PyBullet) and real robotic hardware.
• Develop and deploy algorithms that leverage multimodal sensor data (vision, LiDAR, IMU, tactile, event-based, thermal, etc.) to enhance robotic intelligence and autonomy.
• Collaborate with cross-functional teams to integrate algorithms into robotic platforms and ensure robust real-world performance.
• Work closely with hardware and sensor teams to understand sensor characteristics and translate them into algorithmic advantages.
• Contribute to both system integration and algorithm innovation, validating new sensors through cutting-edge AI benchmarks.

What you should have:

• Master’s or Ph.D. in Computer Science, Robotics, Electrical/Mechanical Engineering, or a related field, with focus on AI, robotics, or embodied intelligence.
• Familiarity with deep learning architectures for embodied AI, with experience in reinforcement learning, imitation learning, or multimodal perception.
• Experience in integrating hardware sensors into robotic systems, with ability to debug both software and hardware pipelines.
• Hands‑on experience with simulation platforms (e.g., Omniverse, MuJoCo, Isaac Sim, PyBullet).
• Proficiency in Python and C/C++, with experience in ROS, PyTorch, or TensorFlow.
• Strong problem‑solving skills, curiosity, self‑motivation, and passion for tackling challenging problems in robotics.
• Strong teamwork and communication skills.

What will help you succeed:

• Experience with C++, CUDA, ROS; publications in top‑tier conferences (CVPR, ICCV, NeurIPS, CoRL) are a strong plus.
• Proven track record in international academic competitions, robotics challenges, or AI contests.
• Industry experience in robotics, AI, or autonomous systems.
• Experience in multi‑modal sensor fusion (vision + tactile, vision + IMU, etc.) and demonstrating measurable performance gains in robotic benchmarks.