About Me

Hello! My name is Xuanbiao Zhu, and I am currently pursuing a Master of Engineering in Computer and Information Science/System Engineering at the University of Pennsylvania. I have a strong interest in backend development, distributed systems, and machine learning.

Previous work

Throughout my academic and professional journey, I have worked on various exciting projects related to Distributed System, Machine Leaning and Deep Learning.

Please check my Portfolio to see details!

Distributed System

As for Distributed System, I developed to a Cloud-based Distributed Search Engine with self-built framework and four high-level components(Crawler, Indexer, Ranker, and Frontend), utilizing a master-slave architecture, consistent hashing for data sharding, and advanced data analytics inspired by Apache Spark, thus successfully crawling and processing 1 million+ web pages.

Furthermore, I also contributed a Distributed Penn Cloud System, a comprehensive cloud application supporting user services, email services, drive services, and monitoring services. I designed a robust key-value store inspired by Google’s BigTable, featuring persistence, replication, and fault tolerance.

Moreover, I developed PennSearch, a high-performance distributed file storage and search system, where I implemented dynamic routing protocols and engineered a Chord-based search system to enhance robustness and scalability. Additionally, I have crafted comprehensive Email and Chat services using C++ and multi-threaded server architectures, ensuring optimal performance and reliability in challenging network conditions.

Machine Learning and Deep Learning

In the realm of machine learning and deep learning, I have tackled projects such as predicting diabetic readmissions using ML/DL models with various techniques, including Gradient Boosted Decision Trees and neural networks, achieving significant accuracy improvements.

Another notable project, ADVENT, involved creating a neural style transfer framework that integrates generative methods with object detection, allowing multiple styles to be applied to different objects within a single image.

Additionally, I developed a robust framework for autonomous UAV navigation using both sample-based methods and reinforcement learning techniques, ensuring efficient and collision-free trajectories in 3D static environments.

I am passionate about leveraging my skills and experiences to solve complex problems and build scalable, efficient systems. Feel free to explore my GitHub.