Computer Vision Engineer

Looq AI builds hardware and software for 3D data capture and reconstruction. The Computer Vision Engineer owns core geometry-stack components, taking algorithms such as SfM, visual-inertial SLAM, bundle adjustment, and dense reconstruction from research into production for on-device and cloud processing.

Responsibilities:

• Design, implement, and improve SfM, SLAM, and visual-inertial odometry for multi-camera rigs with GNSS and IMU
• Improve feature extraction, matching, retrieval, and loop closure. Use classical and learned methods
• Build and tune bundle adjustment and pose graph optimization for large, georeferenced datasets
• Develop dense reconstruction: multi-view stereo, meshing, and Gaussian splatting
• Make the pipeline robust on long-tail field data: low texture, repetitive structure, poor lighting, and sensor faults
• Optimize C++ and CUDA code for GPU and CPU in the cloud and on embedded Jetson hardware
• Define accuracy metrics. Build evaluation tools and benchmarks against survey-grade ground truth
• Evaluate new research, such as feed-forward 3D models, and move the useful parts into production
• Lead design reviews and mentor engineers