London-Based Robotics Research Lead - World Models & 3D AI

Reka; Greater London; On-site; GBP 70,000 - 110,000; Full time

Benefits offered by this job

Visa support

5 weeks paid leave

Health benefits (vision and dental)

Job summary

Reka is seeking researchers to advance robotics through world models, with emphasis on 3D/4D scene representations and multimodal AI. You will work with ML researchers and hardware teams to ensure robust perception, simulation, and control of robotic systems in real-world settings.

The role focuses on building training and simulation platforms, developing persistent scene representations, and pushing the boundaries of generative models for embodied AI.

Qualifications

• Background in robotics or computer vision.
• Experience with 3D perception, scene representations, or world models.
• Experience with robotic systems, simulation, or real-world platforms.
• Strong programming skills (Python, C++, or similar).

Responsibilities

• Build the Foundation: Collaborate with our world model team to build state-of-the-art training and simulation platform for robotics.
• Build the "World": Develop persistent 3D/4D scene representations that maintain temporal consistency.
• Drive Intelligence: Unlock advanced robotics planning and decision making through our in-house, cutting-edge world models.
• Sim2Real: Ensure sensing and system dynamics perform reliably in high-stakes, real-world operations, where models think, simulate, and act.
• Push Boundaries: Partner with top ML researchers to innovate on generative models and physical AGI.

Skills

Robotics

Computer vision

3D perception

World models

Python

C++

ROS

PyTorch

TensorFlow

Tools

ROS

PyTorch

TensorFlow

Job description

Reka is seeking researchers to advance robotics through world models, with emphasis on 3D/4D scene representations and multimodal AI. You will work with ML researchers and hardware teams to ensure robust perception, simulation, and control of robotic systems in real-world settings.

The role focuses on building training and simulation platforms, developing persistent scene representations, and pushing the boundaries of generative models for embodied AI.