Selected Projects

Thanks for these wonderful experiences and supervision!

Agentic RAG System for Code QA

PayPal

CodeRAG is an agentic RAG framework for code understanding that integrates multi-source retrieval and iterative refinement. It preserves code structure through AST-aware chunking and models cross-file relationships using a code knowledge graph, enabling accurate and context-aware reasoning over large code repositories.

AR Laparoscopy Surgery Training System

Computer Aided Medical Procedures & Augmented Reality Lab, TUM
Supervised by Alexander Winkler, Dr. Christian Heiliger (2024)

A novel AR simulation-based training system, which integrates interaction on a 3D phantom and a synchronized 2D augmented laparoscopy visualization system providing three major functionalities: augmented laparoscopic view, interactive phantom operation, and contextual guidance.

DECIDE: Knowledge-Based Version Incompatibility Detection in Deep Learning Stacks

Department of CS, Purdue University
Supervised by Prof. Tianyi Zhang (2023)

An interactive web tool that visualizes a weighted knowledge graph containing 2,376 version knowledge extracted from Stack Overflow discussions. Decide allows the users to easily check whether two libraries are compatible or not, search for a third-party DL component, and shows relevant SO posts as references.

Furniture and Classroom Reconstruction in Virtual Classroom

Department of CS, the University of Hong Kong
Supervised by Prof. Chuan Wu (2023)

An optimized pipeline and container to generate 3D furniture models using single 2D images based on Unicorn and Rembg library. In the system, users can upload furniture images, wall images to create customized 3D virtual classroom.

FoodPrint: An AI-Powered Virtual Diet Assistant IOS App

Department of CS, the University of Hong Kong
Supervised by Dr. T.W. Chim (2023)

Our AI-powered Diet Assistant App, FoodPrint, has revolutionized the way to track and manage fasting routines. With FoodPrint, you can experience the convenience of a dynamic fasting timer, an image-based calorie calculator, a comprehensive diet and weight tracker, and a personalized LLM-powered chatbot diet instructor.

Skills

  • Programming Tools: C/C++/C#, Python, SQL, Java, CSS, JavaScript, PHP, Linux Shell, Swift, React, Node.js, Express.js
  • Software Tools: Android Studio, Bitbucket, Figma, Firebase, Sourcetree, Docker, Django, Unity, Xcode
  • Data Operation Tools: Airflow, Google Analytics, Data Studio, Firebase, MongoDB, Tableau, PostgreSQL