SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: GBP 103,500 - 202,500 / annual
Niantic Spatial is building the future of physical AI, powered by a proprietary database of over 30 billion posed images. The company's mapping technology enables spatial intelligence for robotics, public sector, and energy/industrial markets. The company's reconstruction technology captures environments with geometric accuracy from standard cameras, and its Visual Positioning System delivers precise positioning globally.
The Research Scientist, Embodied AI role sits within Niantic Spatial's R&D team in London, focused on advancing the Large Geospatial Model and pushing the frontiers of Spatial AI. You will work at the intersection of computer vision, machine learning, AI, graphics, and robotics, addressing the sim-to-real gap for visual-spatial understanding in autonomy systems.
Key responsibilities include:
- Developing new spatial representations that embodied AI systems can learn from, train in, and act on
- Advancing real-to-sim pipelines connecting real-world observations, 3D reconstruction, simulation, and robotic policy learning
- Solving emerging spatial intelligence problems related to 3D scene representation, training data, model efficiency, and interactive systems
- Taking new capabilities through to evidence by measuring impact on policy outcomes, not just reconstruction improvements
- Building proof-of-concept prototypes for internal use and production-scale deployment on customer data
- Collaborating with researchers, engineers, and product partners to bring novel models to production
- Publishing and presenting research findings
The role emphasizes discovery, learning, and deploying novel inventions. You will leverage decades of experience from the team's work at Google Maps, Google Earth, and Niantic Labs.
Requirements:
- PhD in computer vision, robotics, graphics, machine learning, or related field, OR equivalent research experience with publication record at top venues
- Substantial expertise in robotics or embodied AI, including sim-to-real transfer, domain randomization, policy evaluation, and simulation-based learning
- Deep experience with 3D scene representations (Gaussian splatting, neural fields, meshes, point clouds, implicit representations)
- Applied experience with generative or diffusion models on 3D data (scene completion, inpainting, relighting, novel view synthesis from sparse capture)
- Experience with 3D semantics, open-vocabulary scene understanding, or scene graphs
- Proficiency in Python and deep learning libraries such as PyTorch
- GPU or graphics programming experience (custom CUDA kernels, shaders, differentiable rendering)
- Track record of contributing to widely used open-source research codebases
Nice to have: C++ proficiency
