A 2-year CMU CS master,
Experienced in scrum software development,
Passionate in backend data-driven problems, resolving ambiguities, taking challenges and making contribution.
Looking for FULL-TIME position in 2016.
Led 5 other interns, built a website based on Struts-Hibernate-Spring (SSH) framework, and utilized project management tools like Git for collaborated work.
After all tokens are labeled, NLU(natural language understanding) then transforms the raw text into structured data, including info like utterance's intent, and its details(we called it Slot). For example, play music intent has slots like song name, artist name, and etc. Slots need to be resolved into entities by searching in remote indexes. My project provides configurable catalog policies for resolution. After my changes, any client could add his customized catalog policies based on his need, and they could help him enhance both search speed and accuracy.
I'm TAing course 11642: Search Engine(Jamie Callan) for Fall 2015 semester. I'm responsible for office hours, grading, design of homework and exam.
Designed and implemented a text-based large scale search engine indexed with Lucene API on corpus of 500,000+ documents from ClueWeb09 dataset Evaluated by metrics like MAP and NDCG. Supported query operators(like NEAR, WINDOW, WSUM) and retrieval models including Ranked Boolean, BM25, Indri and etc. Applied Sequential Dependency Model, pseudo relevance feedback to expand queries and improve performance. Adopted SVM for Learning-to-Rank Algorithm to improve document reranking.
With 3 members, built a BioASQ system among PubMed articles aiming at automatically answer biomedical "yesno" questions, using UIMA framework and third party tools like OpenNLP, Lucene.
Designed system pipelines, including questions generating, document retrieval and then re-ranking, and extract exact answer.
We're building a travel tool to disrupt the traditional way of organizing a trip when we have to spend a lot of time on searching from various websites for the flight tickets, hotels and places to visit, collecting information, and comparing the results to make final decision. With our tool, you can get rid of all these troubles. With just a few clicks, your interest, profiles will be taken into consideration to make reasonable recommendation due to our real-time machine learning techniques.
Aimed to see close relationship among my classmates by analyzing and extracting useful features from their call record data. Applied k-means, where k=2, to group contacts in each call record into clusters. Obtained both labeled training and test data from classmates under professor's support. Finally, visualized the relationship result with graph whose edges are represented by the relation score of every two nodes.
This is a website for users to write and share their blogs. Users could view and post articles and also write comments. Top hot articles are listed on the homepage.
It's a group project on Content-based Cloth Image Retrieval by crawling 20000+ Amazon’s women clothes.
Currently, clothes are searched by text-content. What if we see some fancy clothes in department store and what to search on Amazon? In our project, we could upload an image and search clothes by it. Also, we have a better performance to search clothes in specific color group.
This is an android app to schedule daily courses. After importing your courses, you could view them anytime and anywhere.
Collaborated with 3 teammates, designed a strategic game aimed to help people with healthy diets using HTML5, jQuery and etc. Obtained the Best Participation Award, Imagine Cup, Microsoft’s Global Student Technology Competition (2013).
@Copyright 2015 |Developed by Zhiyue Liu