Computer Vision Engineer with 3D Modeling Skills

DeepX is seeking a highly skilled Senior 3D Computer Vision Engineer with deep expertise in 3D reconstruction pipelines. The ideal candidate will be responsible for designing, developing, and optimizing our core reconstruction engine. You will tackle challenging problems in 3D geometry, multi-view stereo, and large-scale data processing to create high-fidelity 3D models of objects and environments. A key aspect of this role is strong software architecture and coding skills. You will be expected to build a fast, modular pipeline with interchangeable components, allowing us to rapidly iterate and integrate the best-performing algorithms.

Key Responsibilities

• Design, implement, and maintain a high-performance 3D reconstruction pipeline from multi-view imagery.
• Develop and integrate state-of-the-art algorithms for key stages of the pipeline, including feature detection/matching, camera pose estimation, dense matching (MVS), and surface reconstruction.
• Architect the system with a focus on speed, scalability, and modularity, enabling easy substitution of different algorithmic components.
• Work extensively with 3D data, including point clouds and meshes, focusing on generating geometrically accurate and visually faithful reconstructions.
• Apply deep knowledge of projective geometry and 3D transformations to solve complex multi-sensor fusion and real-time mapping challenges.
• Optimize code for performance, including leveraging GPU acceleration where necessary.
• Stay current with the latest academic research and industry trends in SfM, MVS, SLAM, and neural rendering to continuously improve our pipeline.

Required Qualifications & Skills
English proficiency: Upper-Intermediate or higher.

Programming Languages: C++, Python 

Core Libraries & Tools:

• NumPy: For advanced numerical and mathematical operations.
• OpenCV: For fundamental image processing and computer vision tasks.
• Open3D / PCL (Point Cloud Library): For 3D data processing, visualization, and point cloud manipulation.
• PCL: For point cloud processing.
• COLMAP: Practical experience with its SfM/MVS pipeline.
• Plotly / Matplotlib: For 2D and interactive 3D data visualization.
• PyTorch / TensorFlow for GPU-based processing.

Core Technical Skills:

• Projective Geometry: A must-have, with a deep understanding of (camera parameters, projections to/from camera frame, depth-to-disparity conversion, triangulation/trilateration, multi-sensor fusion & real-time mapping.)
• Vector & Plane Transformations: Mastery of rotations (including quaternions), coordinate transformations, and basis conversion for 3D data manipulation.
• SfM & MVS: Proven experience with Structure-from-Motion and Multi-View Stereo pipelines.
• SLAM: Experience with SLAM pipeline components: feature extraction and matching, bundle adjustment, pose optimization, and loop closure.

Preferred Qualifications

• Experience developing a complete 3D reconstruction pipeline from scratch.
• Experience with modern neural rendering techniques such as 3D Gaussian Splatting (3DGS) or NeRFs.
• Knowledge of state-of-the-art feature matching models (e.g., SuperGlue).
• Experience using OpenCV, Open3D, and PCL for camera pose estimation and sparse reconstruction.
• Experience with dense reconstruction techniques, such as Gaussian Splatting and other Multi-View Stereo (MVS) methods.

• Experience with both classical and neural network-based surface reconstruction methods for mesh generation.

   

  About DeepX

  DeepX is an R&D intensive and innovation-driven consortium that provides Artificial Intelligence-powered Computer Vision solutions for businesses. To find out more about us, please visit: https://deepxhub.com/