Research Scientist - 3D Reconstruction (SfM & SLAM)
Spaitial
seniordataother London 91 days ago via Arbeitnow
See how well this job matches your profile
Sign up to get an AI match score and generate a tailored application in seconds.
Get your match scoreTags
3D ReconstructionSfMSLAMMulti-View GeometryBundle AdjustmentNonlinear OptimizationDeep MVSLearned MatchingMonocular Depth EstimationPython
About the role
Role overview
Research Scientist focused on 3D reconstruction (SfM & SLAM) at SpAItial. You will develop methods to recover accurate camera poses and geometry from real-world imagery, combining classical multi-view geometry with state-of-the-art learned reconstructors.
Responsibilities
- Design camera pose estimators and 3D reconstructors.
- Build robust SfM and camera tracking pipelines for multiple imaging sensor types.
- Develop bundle adjusters and nonlinear optimizers, including non-perspective camera formulations.
- Integrate and extend SOTA feed-forward reconstructors (e.g., VGGT, DA3, Pi3).
- Advance deep multi-view stereo, learned matching, and monocular depth methods for dense geometry.
- Create evaluation metrics for pose accuracy and reconstruction quality; improve performance on public and internal benchmarks.
- Scale reconstruction methods to large, diverse real-world datasets while maintaining reliability and efficiency.
- Collaborate with researchers to bring reconstruction advances into production systems.
Requirements
- PhD in computer vision with research focus on 3D reconstruction.
- Publications at top venues (CVPR, ICCV, ECCV, NeurIPS).
- Strong knowledge of multi-view geometry (camera models, epipolar geometry, triangulation, PnP, etc.).
- Familiarity with SOTA deep reconstructors (VGGT, DA3, Pi3) and related areas: deep MVS, learned matching, monocular depth.
- Hands-on experience with SfM/SLAM systems (e.g., COLMAP, ORB-SLAM).
- Experience with nonlinear least-squares solvers (e.g., Ceres).
- Strong Python and PyTorch skills.
- Ability to debug failure cases on challenging real-world data.
Nice-to-haves
- Experience shipping production-grade 3D reconstruction systems.
Location
- London
About Spaitial
SpAItial is building next-generation “world models” that move beyond 2D pixels to physically grounded 3D understanding. The company applies generative AI and computer vision to recover and simulate realistic geometry for industries such as robotics, AR/VR, gaming, and cinema.
Scraped 8/23/2026