Projects

Independently build arXiv parsing and Data-Juicer-based GitHub processing, combining distributed deduplication with human–agent error-driven iteration.

An independently developed inference framework with optimization skills, benchmarks, fused graphs, asynchronous speculation and Rust parallel tokenization.

Core contributions to three-head MTP selection, 2048-token sliding windows and full-graph speculative execution for long contexts.

Core work on lossless parallel tokenization and FlashComm communication/stream adaptation, with LoPT published at ACL 2026.

This program provides a detailed implementation of a fundamental paper in rendering, focusing on Bidirectional Path Tracing (BDPT), Multiple Importance Sampling (MIS). The whole project is written in pure C++ based on the architecture of path tracing assignment from GAMES101 by Prof. Lingqi Yan (UCSB).

This renderer is written in C++ based on OpenGL, which is used for my own Computer Graphic’s study.

It is a software renderer that implements the whole graphics pipeline, including rasterization, shading, and texturing. The renderer is designed to be simple and easy to understand, so it is a good project for beginners to learn computer graphics.