Research Scientist, Embodied AI

DeepRec.ai; Greater London; Hybrid; GBP 120,000 - 200,000; Full time

Job summary

DeepRec.ai in London is hiring a Research Scientist for an Embodied AI group to close the sim-to-real gap. You will work on converting real-world captures into simulation-ready environments for robot training and evaluation.

The role emphasizes inventing algorithms across CV, ML, graphics and robotics, and contributing to a large-scale geospatial model for embodied AI with a strong research ethos.

Qualifications

• PhD in Computer Vision, ML or Robotics.
• First-author publications at CVPR, ICCV, ECCV, NeurIPS, ICLR or similar.
• Strong grounding in 3D vision: SLAM, SfM, neural rendering, Gaussian splatting or similar.
• Excellent research code in PyTorch or JAX.
• Bonus: robotics simulation, sim-to-real, VLMs or policy learning.

Responsibilities

• Inventing new algorithms at the overlap of computer vision, machine learning, graphics and robotics.
• Advancing a large-scale geospatial model for embodied AI use cases.
• 3D reconstruction, scene understanding and spatial reasoning for robots and other machines.
• Taking research from paper to working systems used by real robotics teams.
• Publishing at top venues alongside a strong, research-led team.

Skills

3D vision

Research coding

Education

PhD in Computer Vision, ML or Robotics

Tools

PyTorch

JAX

Job description

Research Scientist - Embodied AI | London (Hybrid)

upto £200,000

DeepRec.ai is partnering exclusively with a well-funded spatial AI company building a foundation model of the physical world, the layer robots and AI agents need to understand real places, not just pixels.

Their London research lab is hiring a Research Scientist into a dedicated Embodied AI group. The focus is closing the sim-to-real gap: turning real-world captures into accurate, simulation-ready environments that robots can be trained and evaluated in.

What you'll work on

• Inventing new algorithms at the overlap of computer vision, machine learning, graphics and robotics
• Advancing a large-scale geospatial model for embodied AI use cases
• 3D reconstruction, scene understanding and spatial reasoning for robots and other machines
• Taking research from paper to working systems used by real robotics teams
• Publishing at top venues alongside a strong, research-led team

What they're looking for

• PhD (or equivalent) in Computer Vision, ML or Robotics
• First-author publications at CVPR, ICCV, ECCV, NeurIPS, ICLR or similar
• Strong grounding in 3D vision: SLAM, SfM, neural rendering, Gaussian splatting or similar
• Excellent research code in PyTorch or JAX
• Bonus: robotics simulation (Isaac Sim, MuJoCo), sim-to-real, VLMs or policy learning

Why it's interesting

• Rare chance to work on embodied AI with real-world data at global scale
• Research-first culture with strong academic links
• Small, senior team where your work ships