About Me
Hey, I am Susie. Nice to Meet you 🤗.
I am currently a full time Software Development Engineer in Amazon. In the year 2019 in May, I graduated at the Univeristy of Illinois at Urbana Champaign with a Master Degree in Computer Engineering. I have experience in multi-thread programming, database(SQL & NoSQL) development, GPU parallel computing optimization as well as other software development engineer's required skills. Before I came to U of I, I had four year work experience in State Grid Corporation of China as a electrical engineer, where I learned dedication and
decision making.
Native AWS Migration Project
- Implemented an end-to-end data pipeline with Spark on Amazon EMR, which process TB-scale data
- Experienced in extract transform and load (ETL) processing large datasets of different forms with AWS Glue
High Performance GPU Computing
- Implemented a baseline many-core parallel algorithm for triangle counting using General Matrix Multiplication
- Optimized the algorithm by GPU specialized sparse matrix application, reducing runtime by 2-3 times
- Optimized the algorithm further by thread coarsening and privatization, accelerated runtime by 25%
- Counted the triangles with 200K nodes within 50s
Simple Distributed File System in Java
Github Source
- Constructed a master server for workload distribution, and membership control with time-bounded completeness
- Designed the system tolerable to failures through a ring-based heart-beating mechanism
- Achieved scalability by balanced workload and efficient bandwidth
Optimization of Convolution Layer in Neuro Network of MXNet for GPU
Github Source
- Converted convolution into matrix multiplication by unrolling input features and filters
- Implemented tiling method for memory reuse, and double buffering to reduce synchronization overhead using CUDA
- Classified 10000 images in 60ms with the speedup of 80 times compared to baseline
Event Search and Recommendation System in Java
Github Source
- Developed an interactive web page for users to search events and purchase tickets (HTML, CSS and JavaScript)
- Improved personalized recommendation based on search history and favorites
- Created Java servlets with RESTful APIs to handle HTTP requests and responses
- Built relational and NoSQL databases (MySQL, MongoDB) to capture real business data
- Designed content-based recommendation algorithms to for business recommendation
- Deployed server to Amazon EC2 to handle 150 queries per second
