SpAItial is pioneering the next generation of World Models, pushing the boundaries of generative AI, computer vision, and the simulation of reality. We are moving beyond 2D pixels to build models that natively understand the physics and geometry of our world. Our mission is to redefine how industries, from robotics and AR/VR to gaming and cinema, generate and interact with physically-grounded 3D environments.
We're seeking a Research Scientist focused on 3D reconstruction. You will advance methods that recover accurate camera poses and geometry from real-world imagery, working with both classical multi-view geometry and state-of-the-art learned reconstructors. The work includes structure-from-motion, bundle adjustment, SLAM, or feed-forward reconstruction, with a focus on robustness, accuracy, and methods that hold up on diverse real-world data.
Responsibilities
Design camera pose estimators and 3D reconstructors.
Build robust SfM and camera tracking pipelines for a variety of input imaging sensors.
Develop bundle adjusters and nonlinear optimizers, including non-perspective camera formulations.
Integrate and extend SOTA feed-forward reconstructors (VGGT, DA3, Pi3)
Advance deep multi-view stereo, learned matching, and monocular depth methods for dense geometry.
Build evaluation metrics for pose accuracy and reconstruction quality, and drive improvements against public & internal benchmarks.
Scale reconstruction methods to large, diverse real-world datasets while keeping them reliable and efficient.
Collaborate with researchers to bring reconstruction advances into production systems.
Key Qualifications
PhD in computer vision with a research focus on 3D reconstruction; publications at top venues (CVPR, ICCV, ECCV, NeurIPS).
Deep understanding of multi-view geometry: camera models, epipolar geometry, triangulation, PnP, etc.
Strong familiarity with SOTA deep reconstructors (VGGT, DA3, Pi3) and related areas such as deep MVS, learned matching, and monocular depth estimation.
Hands-on experience with SfM/SLAM systems (COLMAP, ORB-SLAM) and nonlinear least-squares solvers (Ceres).
Experience shipping production-grade 3D reconstruction systems is a strong plus.
Strong Python and PyTorch skills.
Comfortable debugging failure cases on challenging real-world data.
At SpAItial, we are committed to creating a diverse and inclusive workplace. We welcome applications from people of all backgrounds, experiences, and perspectives. We are an equal opportunity employer and ensure all candidates are treated fairly throughout the recruitment process.
Skills Required
- PhD in computer vision with a research focus on 3D reconstruction
- Publications at top venues such as CVPR, ICCV, ECCV, or NeurIPS
- Deep understanding of multi-view geometry, including camera models, epipolar geometry, triangulation, and PnP
- Strong familiarity with state-of-the-art deep reconstructors such as VGGT, DA3, and Pi3
- Knowledge of deep multi-view stereo, learned matching, and monocular depth estimation
- Hands-on experience with SfM and SLAM systems, including COLMAP and ORB-SLAM
- Experience with nonlinear least-squares solvers such as Ceres
- Strong Python and PyTorch skills
- Ability to debug failure cases on challenging real-world data
- Experience shipping production-grade 3D reconstruction systems
What We Do
SpAItial is pioneering Spatial Foundation Models (SFMs), a groundbreaking AI paradigm designed to generate and reason about the appearance and physics of real and imagined environments. SFMs possess an intrinsic understanding of space-time, enabling transformative shifts in applications at the intersection of virtual and physical worlds. Unlike existing generative AI technologies such as LLMs, image, or video models, SFMs operate natively in physical space. This significantly advances their cognitive capabilities, which mimics human understanding. SFMs promise to revolutionize various applications across industries, from creating immersive virtual worlds for gaming and entertainment, to advancing CAD engineering and construction, to powering next-generation VR/AR experiences, and enabling sophisticated, physically-intelligent robotics.
