Python, Java, JavaScript
Resume:
New York, NY, USA
Beijing, China
Beijing
HENGZHI DING
Summary
Software Engineer with 2 years in analysis, design, development, testing and implementation of various applications, experienced in algorithm design and deep learning. Key Skills: Java, Python, JavaScript, PHP, Matlab, MySQL. Education
Master of Science: Electrical Engineering 3.4/4.0
Columbia University
Bachelor of Science: Control Technology
Beihang University
B.S. in Applied Mathematics (dual degree)
Professional and Project Experience
Software Engineer Intern
Neusoft
Judge whether a picture generated by X-ray machine contains contrabands (knife or guns), by implementing a convolutional neural network, the accuracy rate can reach up to 90%. Used faster-rcnn method (object detection) to detect knives and guns in X-ray machine's pictures. Assisted coding Lee algorithm to realize image enhancement. Application developement
A chatting application. Users can easily register, add friends, chat with each other and receive offline message through user interface.
Based on GBN protocol, periodically calculate link weights of network, update route table by implementing distance vector algorithm.
Bookstore management application. User can search books, add or delete books and modify book's properties.
Websites blocker application, can block any websites from visiting at any period of a day. Web crawler program. Users can get relevant information by specifying key words and target website. Multi-digit-number-guess game, players can get feedback when they make a guess to help them win. A quick dialer application, mainly for aged people. For they may have some difficulty using mobile, the application simplifies operations for them.
Web development
E-commerce store. Users can register, scan information and pay for or store products, etc. Forum website. Users can register, post or reply topics. Reaction game. Random shape appears on the web page with random location, player clicks it to make it disappear, then gets the “reaction” time.
Image process based on deep learning
Based on VGG19 network, built a network which: 1) can combine one picture's content with another picture's art style to create a new picture; 2) user can create image by giving a rough “semantic” layer of his desired picture.
Built a convolutional neural network, which can effectively recover a seriously corrupted picture with 30% content lost.
Successfully reconstructed images using PCA and gradient descent. Activities and Honors
2016 "Tesla Scholar" award when admitted by Columbia University (10%). 2013 and 2014 two-time second-class department academic scholarship (30%). 2012 table tennis department champion.
362 Riverside Drive Manhattan, New York, NY 10025 C: 1-646-***-**** ******@********.***
Contact this candidate