MENGZE ZHU
(510) 646-7039 | mengze_zhu@berkeley.edu
EDUCATION
University of California, Berkeley | M.Eng., EECS Berkeley, CA | Aug. 2019 – May 2020 (Expected)
• GPA: 3.82/4.00
• Coursework: Principles and Techniques of Data Science, User Interface Design and Development, CV
Peking University | B.S., CS Beijing, China | Sept. 2015 – July 2019
• GPA: 3.84/4.00 (Rank: 3/207)
• Related Coursework: Data Structure and Algorithm, Algorithm Design and Analysis, Computer Architecture, OS, Introduction and Experiments of Microelectronics and Circuits, Design and Experiments of Digital Logic
• Awards: National Scholarship (Top 1%), Outstanding Graduates (Top 5%), “Top-Notch Class” Program (Top 5%), First Prize in the 34th National College Student Physics Competition (Top 5%)
SKILLS
• Programming Languages: C /C++ (Proficient), Java, Python, Latex, Verilog, SQL, JavaScript, HTML
• Tools and Frameworks: Pandas, STL, Android Studio, Keras, TensorFlow, Pytorch, Hadoop, Git, Linux
PROFESSIONAL EXPERIENCES
Digital Mobility, Siemens Ltd., China Beijing, China
Data Analyst Intern Sept. 2018 – Mar. 2019
• Streamlined and modularized the workflow of metro transportation data cleaning and preprocessing in Python, leading to a 10x increase in efficiency
• Built an analysis algorithm (a perceptron model) for monitoring bogie health conditions based on public transit operations and energy consumption data
• Used Pandas to categorize metro data and draw related plots for data analysis, detecting metro operation faults and improving passenger travel safety
RESEARCH EXPERIENCES
School of Computer Science, Carnegie Mellon University Pittsburgh, PA
Research Assistant (Advisor: Prof. Katerina Fragkiadaki) July 2018 – Sept. 2018
• Directed a robot’s self-taught exploration of specific tasks (such as playing with toys) by extracting semantic information based on visual and audio information using unsupervised learning
• Introduced a novel audiovisual correspondence learning task, extracting audio feature representation by using deep learning method with greater than 70% accuracy
• Improved audio feature learning in terms of convergence speed by integrating ResNet block into the CNN model
Key Lab of High-Confidence Software Technologies, Peking University Beijing, China
Research Assistant (Advisor: Prof. Xuanzhe Liu) May 2017 – May 2019
• Designed a principled cache for deep learning inference in continuous mobile vision by exploiting temporal locality in input video streams, saving execution time by 18% and energy consumption by 20%
• Developed an innovative lightweight region matcher to improve frame processing efficiency up to 72% by searching for neighboring fine-grained regions guided by video motion heuristics
• Used Android Studio to develop a dashboard including a demo of using the proposed optimized algorithms for object detection in the recorded videos for user experience on android device
• Proposed a deep learning framework for wearable devices to improve performance (up to 23x faster and 86% more energy efficient) by offloading deep learning tasks to an external micro-device
ACADEMIC PROJECTS
Android Project Development Berkeley, CA | Aug. 2019 - Dec. 2019
• Designed and developed electric vehicle charger digital companion app for home charging station
• Designed and developed a mobile app that helps navigate electric vehicle owners to the best charging point.
• Designed and developed Spare Change, a personal finance education app for high school students, letting users to make financial decisions, manage their personal budget and receive tips from experienced people.
Series of Hadoop Projects Beijing, China | Feb. 2019 – June 2019
• HDFS: Built a HDFS for storing and accessing big volume alum
• MapReduce: Used MapReduce to calculate the total revenue of a company's high-scoring movies
• HBase/Hive/Pig: Imported Titanic dataset into HBase/Hive/Pig and implement designed query functions
• Zookeeper: Implemented shared locks to solve readers-writers problem in a distributed environment
PUBLICATIONS
• Mengwei Xu, Mengze Zhu, et al. “DeepCache: Principled Cache for Mobile Deep Vision,” 24th Annual International Conference on Mobile Computing and Networking (MobiCom18), accepted.
• Mengwei Xu, Feng Qian, Mengze Zhu, et al. “DeepWear: Enabling Deep Learning on Wearable Devices via Adaptive Local Offloading,” IEEE Transactions on Mobile Computing (TMC), accepted.
个人简历:
2015-2019 北京大学信息科学技术学院计算机科学与技术系本科
2019-2020 UC Berkeley, Master of Engineering in EECS
