ZEJUN LIU
240-***-**** adkc3l@r.postjobfree.com linkedin.com/in/zejun-liu github.com/ZejunLiu0
EDUCATION
University of Maryland, College Park Sep. 2019 { Dec. 2021 Master of Science in Computer Science GPA: 3.96
Beijing University of Posts and Telecommunications, China Sep. 2015 { Jun. 2019 Bachelor of Engineering in Computer Science and Technology SELECTED PROJECTS
Squirtle: Scalable Distributed Steam Processing Engine with FaaS Services Nov. 2020 - Present
Building a distributed SQL query engine for stream processing on AWS Lambda with Rust, which supports lter, aggre- gation, projection, sort, limit, and join operations.
Adding di erent data sources such as Amazon Kinesis Data Streams and Kafka to the stream processing engine.
Modi ed Apache Arrow source code to enable data batches to ow across Lambda functions. Automatically generated Lambda function code from the client’s SQL query input.
Comparing the engine’s performance with Amazon Kinesis Data Analytics(Flink) and Spark Streaming with the Nexmark benchmark.
UbiStore: A Distributed And Cloud-Based Storage System Source code Sep. 2020 - Dec. 2020
Built a full-scale reliable, highly-available, and secure distributed key-value storage system with Golang.
Implemented version control with the Rabin-Karp algorithm for le chunking, Merkle tree/hash tree for le synchronization, and HTTP and REST APIs for networking.
Enabled encryption with RSA private key and public key. Implemented di erent server security levels.
Implemented a le system on top of the key-value store with the Bazil FUSE interface, which supports le creations, reads, writes, directory creates, and deletions. Files are cached in RAM and made persistent asynchronously and periodically to improve system performance.
Implemented the Raft consensus algorithm, which supports leader election, log replication, and fault tolerance, etc. to replicate les across servers.
VidShare: A Video Sharing Website Github Jul. 2020 - Oct. 2020
Developed a Youtube-like website with MongoDB, Express, React, and Node.js, which supports login, register, upload videos, watch videos, comment, subscribe, and so on.
Designed the front-end with React Bootstrap. Connected the front-end and the back-end through the axios API. Hosted MongoDB on AWS to perform CRUD operations.
Citibike Data Visualizer Github Oct. 2019 - Dec. 2019
Developed a visualization tool for the Citibike dataset, which includes over 1 million bike riding records.
Designed the frontend with Vue.js, Bootstrap. Implemented a bike station map with Mapbox GL JS API and custom design. Implemented several visualizations with Google Charts API.
Built the backend with Flask. Used axios to connect the frontend and backend. Used PostgreSQL to store the records and used Python (psycopg2) to perform CURD operations. GeekOS: A Mini Operating System Sep. 2019 - Nov. 2019
Implemented basic functionalities and components of a 32-bit operating system in C, including system calls, signals, Per- CPU variables, paging, virtual memory, and le system.
Implemented the project on a Vagrant Ubuntu virtual machine environment. Used The GNU Debugger (GDB) to debug the project.
WORK EXPERIENCE
Research Intern Feb. 2018 - Jun. 2018
Megvii Co., Ltd. Beijing, China
Designed a video preprocessing pipeline that processes terabyte-level video data using the knowledge of image processing with Python, Numpy, OpenCV.
Wrote Python scripts to de-serialize data in pickle les and transform them into the format used by the company for model training. Wrote Shell scripts to execute Python programs.
Designed a method to cluster similar images out of thousands (1000 to 9000) of images per video by using knowledge of clustering algorithms, matrix operations with Numpy and OpenCV. SKILLS
Languages C/C++, Java, Python, Rust, Go, HTML, CSS, JavaScript, SQL, Shell Web Development React, Express.js, Node.js, Vue.js, Flask, Django, Bootstrap Database & Tools MySQL, MongoDB, PostgreSQL, AWS, GCP, Linux, Git, Pytorch, Numpy, pandas