42dot
SLAM & 3D Reconstruction Engineer (Autonomous Driving)
Pangyo (Software Dream Center), South Korea, Remote. Posted a month ago
The role
The SLAM & 3D Reconstruction Engineer develops Offline SLAM, Localization, Self-Calibration, and 3D Reconstruction algorithms. They use Multi-view Geometry and Neural Rendering technologies to create high-quality 3D data for AI training.
What you will do
- Develop Offline SLAM and Localization algorithms using Camera, LiDAR, IMU, and GNSS sensor data.
- Develop Trajectory Estimation, Pose Optimization, and Global Alignment algorithms for large-scale driving data.
- Develop Multi-Sensor Self-Calibration and Calibration algorithms.
- Improve 3D Reconstruction accuracy using Multi-view Geometry, Structure-from-Motion, and Bundle Adjustment.
- Research and apply Neural Rendering based 3D Reconstruction technologies like NeRF and 3D Gaussian Splatting to data pipelines.
- Build and improve quality evaluation and verification systems for Localization, Calibration, and 3D Geometry data.
- Apply algorithms to large-scale data processing pipelines and build automated data generation and quality improvement processes.
- Collaborate with Perception, AI, and Data Platform teams to improve AI training data quality.
Required qualifications
- Master's degree in Computer Science, Electronic Engineering, Mechanical Engineering, Robotics, or related fields with 3+ years of experience or a PhD
- Understanding of Computer Vision, Robotics, SLAM, Localization, or 3D Geometry
- Experience in C++ based algorithm development and software implementation
- Experience in data analysis and algorithm verification using Python
- Understanding of Linear Algebra, Multi-view Geometry, Probability, and Nonlinear Optimization
- Development experience in a Linux environment
Nice to have
- Experience developing Visual SLAM, LiDAR SLAM, Visual Odometry, or Offline Localization algorithms
- Experience developing Multi-Sensor Localization, Sensor Fusion, and Sensor Self-Calibration algorithms
- Experience developing Extrinsic, Intrinsic, or Temporal Calibration for Camera, LiDAR, and IMU
- Experience with geometry-based optimization algorithms such as Factor Graph Optimization, Pose Graph Optimization, and Bundle Adjustment
- Experience with Structure-from-Motion (SfM), Multi-view Stereo (MVS), Point Cloud Registration, or Dense 3D Reconstruction
- Experience developing Neural Rendering based 3D Reconstruction technologies like NeRF and 3D Gaussian Splatting
- Experience in Camera and LiDAR based 3D Geometry Reconstruction and quality improvement
- Experience using frameworks such as Ceres Solver, GTSAM, COLMAP, or Open3D
- Experience in large-scale driving data processing and data quality evaluation
- Experience publishing papers in CVPR, ICCV, ECCV, ICRA, IROS, RSS, SIGGRAPH or contributing to open source projects
Before you apply
- A 3-month probation period may apply.
- Reference checks may be conducted with the candidate's consent after the interview process.
- Employment may be canceled if false information is found in the application.
About the company
42dot's 3D Vision Team researches and develops technologies to create high-quality location information and 3D spatial data for Autonomous Driving AI.