- State-of-the-art automated point cloud classification and segmentation of real-world, customer-acquired LiDAR datasets with artificial intelligence- Gaussian Splat training and rendering pipeline for real-world customer-acquired scenes with LiDAR initialization- Sub-centimeter level accuracy factor-graph based localization suite for both GNSS-rich (tightly coupled IMU + GNSS) and GNSS-denied (LiDAR-based SLAM) environments- Adaptive LiDAR-based point cloud moving-object detection- Survey-grade centimeter resolution orthomosaic generation via novel SfM-less pipeline - Robust supervised and unsupervised (automated) algorithms for intrinsic and extrinsic calibration of sensors (LiDAR, IMU, camera), including fully automated LiDAR-to-camera calibration- Point cloud colorization from occluded multi-view scenes- Adaptive point cloud smoothing and denoising- GIS processing suite for survey-grade deliverables generated from user-flown LiDAR data, including high-resolution ground surface model

Independently investigated applying deep learning and AI on remotely sensed and GIS data (e.g. LiDAR-based 3D point clouds and satellite imagery) with Dr. Qiu Fang, GIS department head.Developed novel techniques to process extremely large-scale point clouds of airborne LiDAR surveys with state-of-the-art deep learning algorithms to perform semantic segmentation of ground, building, vegetation, bridge, and power-line points.

5-star freelancer on Upwork assisting with clients from industries including game development, insurance, and remote sensing.

Assisted in server-side and client-side development of background-screening platform.