Responsibilities:
- Lead the design and evolution of our localization architecture, making critical build-vs-buy decisions and defining long-term technical roadmaps.
- Design and implement robust, real-time algorithms for sensor fusion, visual/LiDAR odometry, and global localization using techniques like Factor Graphs, EKF/UKF, and particle filters.
- Develop pipelines to generate, maintain, and query High-Definition (HD) maps.
- Work on "priors-based" localization and semantic map matching.
- Solve the "stale map" problem by designing systems for change detection. You will use fleet data to identify construction zones, lane shifts, or new signage and update the map in near real-time. Maintain and continuously update large-scale LiDAR-based HD maps to support autonomous driving systems.
- Optimize code for embedded compute constraints (latency, memory, and power) without sacrificing safety or accuracy.
- Provide self diagnosis capability for localization
- Act as a technical mentor to senior and junior engineers, conducting rigorous code reviews and fostering a culture of engineering excellence.
Required Skills:
- M.S. or Ph.D. in Robotics, Computer Science, Electrical Engineering, or a related field.
- 7+ years of industry experience in robotics, with at least 4 years dedicated to production-level localization or mapping systems.
- Expert-level Modern C++ (14/17/20) skills, with a deep understanding of memory management, multi-threading, and real-time systems.
- Strong theoretical foundation in probabilistic robotics, including Kalman Filtering (EKF, UKF), nonlinear optimization (Ceres, g2o, GTSAM), and SLAM.
- Deep understanding of 3D geometry, linear algebra, and coordinate transformations (quaternions, SE(3)).
- Familiarity with Eigen, Ceres, PCL, OpenCV, GTSAM
Preferred Skills:
- Experience shipping localization stacks for deployed autonomous vehicles or mobile robots.
- Hands-on experience with LOAM, lego-LOAM, ORB-SLAM, or VINS.
- Experience with CUDA/GPU programming or optimization for ARM-based embedded platforms.
