I'm currently a Master student majoring in Computer Science in Boston University. Interested in full-stack development and Java development, I love to play with new technology and build interesting projects!
• Developed a reading website using React and immutable.js containing over 200 articles on various topics. Used Redux for reducers and redux-thunk as middlewares.
• Implemented login/register interface and login permission check.
• Developed an ordering system with Spring Boot and Vue.js to allow internal employees to order food online.
• Designed class interfaces with Singleton and Factory Method software design patterns.
• Implemented MVC backend using Java. Used Spring Boot for auto-configuration and dependency injection.
• Wrote unit test with Junit4. Automated system deployment project using Nginx.
• Created Java servlets with RESTful APIs to handle HTTP requests and responses written in JSON/XML format and utilized MySQL to capture over 10k event data (name, description, location) from TicketMaster API.
• Designed recommendation algorithms (e.g. content-based recommendation) to improve event recommendation performance by 50% based on search history and favorite records.
• Deployed server to Amazon EC2 to handle multiple queries (around 150) per second tested by Apache JMeter.
• Built a scalable web service in Go to handle posts and deployed to Google Cloud (GAE flex) for better scaling.
• Utilized ElasticSearch(GCE) to provide geo-location based search functions such that users can search nearby posts within a distance (e.g. 200km)
• Used Google Dataflow to implement a daily dump of posts to BigQuery table for offline analysis.
• Aggregated the data at the post level and user level to improve the keyword based spam detection (BigQuery).
• Analyzed and crawled the data set, and using python, pandas, matplotlib, numpy processed the data to filter out the streets that are not suitable for renaming. Stored data utilizing MongoDB.
• Sorted streets according to traffic flow, street length, number of re-names and used z3, k-means to increasing accuracy by 30%.