Christopher Lee

 chris@clee.ai    clee.ai    github.com/clee-ai    linkedin.com/in/christopher-lee-utd

Robotics and AI Engineer with 4+ years of experience deploying full-stack perception systems, transforming multi-sensor data into robust localization, 3D reconstruction, and AI-powered understanding for autonomous platforms.

 Work Experience

• Architected and deployed state-of-the-art AI models for automated classification of real-world, terabyte-scale LiDAR datasets.
• Engineered and maintained a full-stack customer-facing robotics pipeline, from backend C++ algorithm development to frontend interfaces in JavaScript/Electron.
• Drove significant performance and efficiency improvements for the core AI and SLAM algorithms, optimizing C++ and CUDA code for reduced latency and memory usage in constrained production environments.
• Developed a sub-centimeter accuracy localization suite using an EKF with LiDAR SLAM in GNSS-denied environments and a factor-graph optimizer for tightly-coupled IMU/GNSS.
• Engineered a complete training and rendering pipeline for Gaussian Splatting, enabling high-fidelity 3D reconstructions of real-world scenes from vision and LiDAR data.
• Implemented a robust suite of classical computer vision solutions, including automated multi-sensor calibration, SfM-based reconstruction pipelines, and real-time dynamic object filtering.

• Pioneered a novel deep learning methodology for segmenting massive aerial LiDAR datasets, establishing research collaborations with both the Head of the GIS Department and external utility companies.
• Engineered and validated an end-to-end AI pipeline to automatically classify power infrastructure on industry-provided LiDAR scans, developing custom techniques to overcome terabyte-scale data and GPU memory challenges.

 Open-Source Contributions

• Drove core performance improvements by implementing CUDA-level optimizations, adding mixed-precision support, and contributing to features that achieved state-of-the-art inference speedups.
• As a maintainer, ensured library stability and ease-of-use by managing PyTorch API compatibility, authoring user-facing documentation, and resolving community-reported issues.

• Architected a compatibility layer for multiple sparse backends (TorchSparse, MinkowskiEngine), expanded capabilities by adding new models (ex. SPVCNN), and significantly improved memory usage by implementing mixed-precision support.
• As a core maintainer, improved code quality and usability through extensive refactoring, documentation updates, training configuration improvements, and community support.

 Skills

• Programming & Libraries: C++, Python, CUDA, Pytorch, OpenCV, PCL, Open3D, Ceres, GTSAM, scikit-learn, JavaScript
• AI & Computer Vision: 2D/3D Semantic Segmentation, Object Detection, NeRF, Gaussian Splats, SfM/MVS, Sensor Fusion
• Robotics & 3D Data: SLAM, Navigation, EKF, FGO, GNSS, Sensor Calibration, State Estimation, Point Cloud Processing
• Tools & Platforms: AWS (EC2), Docker, ROS, Git, CMake, vcpkg, Wandb, TensorBoard, Electron, Node.js

 Patents

Knoll, Alexander R.; Knoll, Harrison L.; Lee, Christopher; Whitney, Timothy; & Stanislas, Leo N. (2023). Processing LiDAR Data. U.S. Patent Application No. 18/512,881, filed November 17, 2023. Patent Pending.

 Education

Bachelor of Science, Computer Science

• Award: 2020-2021 University of Texas at Dallas Undergraduate Research Scholar

 Community Service

• Provided care for shelter cats and dogs, from daily walks and kennel maintenance to implementing targeted play and socialization strategies designed to make less sociable animals more adoptable.