Research Scientist - 3D Reconstruction (SfM & SLAM)
Core
Designing and building robust 3D reconstruction systems (SfM, SLAM, bundle adjustment) to recover camera poses and geometry from real-world imagery for robotics, AR/VR, gaming, and cinema.
Role type
Research Scientist (3D Reconstruction)
Builds
Production-grade 3D reconstruction pipelines and evaluation metrics
Domain
Computer Vision / 3D Geometry / Generative AI
Deliverable
production ML models
Required skills
Multi-view geometry, SfM/SLAM systems, nonlinear optimizers, deep multi-view stereo, learned matching, monocular depth estimation, Python, PyTorch
Preferred skills
Experience shipping production 3D systems, debugging challenging real-world data
Technologies
COLMAP, ORB-SLAM, Ceres, VGGT, DA3, Pi3
Responsibilities
Design camera pose estimators and 3D reconstructors; Build robust SfM and camera tracking pipelines; Develop bundle adjusters and nonlinear optimizers; Integrate and extend SOTA feed-forward reconstructors; Advance deep multi-view stereo and learned matching methods; Build evaluation metrics for pose accuracy and reconstruction quality; Scale reconstruction methods to large, diverse real-world datasets; Collaborate with researchers to bring advances into production systems
Seniority
Senior, hands-on IC
