About Me
My name is Zui Tao, in Chinese it's written as 陶醉, you can also call me Steven. I am currently a graduate student at Information Networking Institute, Carnegie Mellon University. I am interested in cloud PaaS and other networking backend service. I want to build things that people really care.
Experiences
My job at Intel was mainly testing and evaluating the Intel-built, open-source cloud platform named Trusted Analytics Platform(TAP).
TAP is designed for programmers and data scientists to accelerate the creation of advanced analytics and machine learning solutions. TAP is built based on the Cloudera as the data engine and CloudFoundry as the resource engine. The source code of TAP can be found on Github.
My undergraduate research with Dr. Weikang Qian was focused on approximate computation and synthesizing approximate circuits.
The goal is to simplify a digital circuit given the error rate threshold. The algorithm utilizes the signal probability based cut and extended don't care terms. ABC was used to implement and verify the algorithm.
Projects
- Cooperated with Intel to design and implement a real-time face detection algorithm
- Applied sliding windows and image pyramid to preprocess each frame of the video
- Limited execution time per frame to 40ms using Cifar-10 Neural Network under Caffe framework
- Implemented MFCCs on Matlab to extract the unique audio features of a person’s speech
- Trained the datasets using support vector machine and neural network in Matlab to reach an error rate at 9%