• Agentic system in the game engine

• Character/Avatar physically-based rendering performance & visual quality improvements

• Gaussian Splatting-based Codec Avatar lighting

• Codec Avatar <> game engine integration

Worked on maintaining and developing world-scale aerial & ground-level photogrammetry 3D reconstruction pipeline written in C++.

• Led and launched a quality benchmarking system that measures the aerial reconstructed data quality using photo-consistency metrics, providing a direct way to catch regression and track algorithmic improvements; Collaborated with data scientists to analyze the quality metrics of Google's worldwide 3D data.

• Explored the new capability of generating semantic segmented mesh products from labeled imagery; Implemented visualization support in glTF and other in-house 3D geometry formats.

• Improved the pipeline infrastructure: simplified pipeline stage dependency graph, reduced pipeline latency, and resource usage, improved integration test coverage, troubleshoot production issues, etc.

Maps and Geospatial

• Bing Map 3D Reconstruction Pipeline (C++, C#, Azure)

Teaching Assistant of CS180 Computer Graphics (Spring 2019)

Teaching Assistant of CS165A Artificial Intelligence (Winter 2019)

Helped two young faculties build their first undergrad course.