Zhihao(Zach) Deng


image

  zacdeng0720@gmail.com   Github   My Blog   Linkedin


Education Background

Northwestern University (NU), US 09/2021 - 12/2022

  • Master of Science in Computer Engineering, GPA: 3.764/4.0

Northeastern University (NEU), CN 08/2017 - 06/2021

  • Bachelor of Engineering in Robot Engineering, GPA: 3.473/5.0

  • Serve as the president on Student Union of Faculty of Robot Science and Engineering, NEU


Proficient Skills

Programming language:

  • Typescript, Python, MSSQL, C#, Javascript, Java, GraphQL, HTML, CSS(SASS), C/C++, Bash/Shell, MATLAB

Developer Tools:

  • AWS, Postman, Git, Datadog, Apollo, Figma, Docker, Heroku, Firebase

Database:

  • Microsoft SQL Server, DynamoDB, MySQL, SQLite3, NoSQL

Web Technologies:

  • Angular, GraphQL, ReactJS, Redux, Node.js, Flask, Django, Bootstrap, RESTful API

Other:

  • ROS, Linux, OpenCV, Pandas, Markdown, Jira, PyTorch, Jupyter, Anaconda, Tensorflow, Latex, Labview, Json

Work Experience

NRG Kiosk LLC (Power Kiosk) - Full Stack Software Engineer - 01/2023-present

  • Participated in all phases of the agile software development life cycle as a full stack developer, including frontend(Angular), backend(Controller-Service-Repository pattern), GraphQL middleware and database(MSSQL).

  • Developed a C# rate matrix loader for multiple suppliers, extracted email attachments from AWS S3, and engineered RESTful APIs to retrieve constant variables. Integrated the results into database. Configured AWS CodePipeline and EC2 instance, and established a CronJob to automate this as a daily routine.

  • Tuned stored procedures to promote reusability of code, thereby improving system performance and reliability. Monitored Datadog and Apollo server performance to ensure smooth operation and conducted routine maintenance.

  • Analyzed and implemented the contract reinstatement process, including database design and tuning, RESTful APIs development, frontend and backend logic implementation. Helped the company retain customers, bringing in higher profits.

  • Designed a Python-based OCR model that automatically verifies and processes billing files, deployed on AWS EC2 with a scheduled CronJob. This automation significantly enhanced system efficiency by minimizing the need for manual verification.

Healthy Hop LLC - Full Stack Software Engineering Intern - 05/2022-09/2022

  • Partnered with UI/UX team to design mock-ups and prototypes of the components on Figma. Utilized ReactJS and NodeJS to implement the frontend and backend logic. Experienced in DynamoDB tables design using NoSQL.

  • Designed RESTful APIs using AWS gateway and conducted testing using Postman, implemented the lambda functions with NodeJS to interact with the DynamoDB.

  • Optimized UI rendering speed by 25% using debouncing in react and limited excessive API requests from the text field. Engaged in front-end test and lifted test coverage by 40%. Collaborated with systems team to implement CI/CD pipelines.

  • Refactored product code to fix code smells and certain structures that indicate violation of fundamental design principles.

iFlytek, Hardware dept. - Software Development Engineer - 01/2021-05/2021

  • Revamped the remote-control module for driverless car in the web application to improve the control precision. Designed and implemented user-friendly components for real-time map building and path planning visualization using Flask, CSS and Javascript.

  • Migrated the C-based navigation and path planning algorithms for driverless car using Python. Developed the backend of the application using Flask framework

  • Collaborated with Tsinghua University to offer training for 34 ROS beginners in speech recognition,map building, localization, and navigation.


Research and Projects

1. Research about Reinforcement Learning Based on Google Research Football Framework - 09/2020 06/2021

Member of the Research Lab (Tutor: Zixi Jia, Associate Dean of Faculty of Robot Science and Engineering, NEU)

  • Researched about reinforcement learning in multi-agent system under the instruction of Zixi Jia and Jingyu Ru.

  • Designed a multi-agents football simulation with Unity3D ML-Agents and used C# to add specific features.

  • Proposed a distributed PPO algorithm, which uses multi-threaded parallel computing to significantly shorten the training time by 22 percent

  • Proposed a strategy algorithm that uses the real-time scores of the game as a penalty reward when training the model. The trained model based on this algorithm greatly improves the winning rate of the agent by 20 percent in most scenarios

2. A Racing Smart Car based on Deep Learning (Sponsored by Baidu) - 10/2019 - 08/2020

Co researcher & Team Leader (Tutor: Prof. Zheng Fang, NEU)

  • Completed the PID control program for the smart car mainly with Python. Enabled the smart car to avoid obstacles independently with a monocular camera using OpenCV

  • Equipped the smart car with the function of collecting the data of relevant marks, conducted CNN under the Paddle framework, identified obstacles and marks at high speed, increased the identify accuracy by 30%

  • Led the team to take part in the 15th National College Students Smart Car Competition and won National First Prize

3. Verification Code Recognition Based on Convolutional Neural Network - 04/2020 - 06/2020

Team Leader (Tutor: Prof. Nick Feamster, University of Chicago)

  • Under the requirement of applying machine learning to solve a real world problem, my team decided to conduct this project

  • Randomly selected four from 56 numbers and symbols as the verification code with Captcha Library in PyPI to solve the verification code problem in crawling

  • Set up a CNN network with multiple convolutional layers to train characters so that verification codes comprised of characters could be automatically recognized in the crawling process without human intervention

  • Constructed environments in Linux and Tensorflow, designed and set up the structure of the convolutional neural network, adjusted parameters, and completed training and experiments

  • The final recognition accuracy rate was 92%, and the project was rated A. First author of an abstract paper on Conference on Soft Computing & Machine Intelligence (ISCMI 2020)

4. An Intelligent Rubik’s Cube Resolving Robot - 11/2019 - 02/2020

Team Member (Tutor: Researcher Shuai Wang)

  • Teamed up with other students to conduct this project for participating in the Provincial Contest of the National Artificial Intelligence Competition

  • The competition required contestants to enable robots to revert cube based on stm32 and computer vision

  • Constructed the mechanical structure of the robot, dealt with cube reverting algorithms, used Keil and C/C++ to optimize the algorithm as well as the PID control of the motor, and applied STM32CubeMX to set the single chip microcomputer

  • The completed robot could revert a magic cube to its original state within 20 seconds

5. OpenCV Based Probot Anno Arm Gripping - 09/2019 - 01/2020

Researcher (Tutor: Researcher Shuai Wang)

  • Proposed a ROS-based robot arm control system design and aimed to solve the difficulty in traditional manipulator control system design and the poor portability

  • Added binocular vision based on traditional robotic arm control to avoid obstacles and grab objects

  • Used MoveIt! to plan the trajectory in a point to point manner and displayed it dynamically on Rviz

  • Implemented binocular vision based on OpenCV

  • Completed camera calibration, image acquisition, preprocessing, feature extraction, etc., calculated the target distance based on the parallax value, and fed back to MoveIt!

  • Conducted independent planning through MoveIt! and completed grab movements

6. Subscript Web Application 09/2021 - 09/2021 - 10/2021

  • Designed and established a web application with classmates for managing subscriptions in software development class

  • Used React to set up the app, styled the app by JavaScript inline, CSS and Material UI. Achieved the user authentication using Google Sign-In and interaction with database, finished user-related functions with React Hook

  • Made it public using Firebase, used Cypress to test the web page and writing a series of unit tests by Jest

7. Recovery Tracker Software System (Full Stack) - 02/2022 - 03/2022

  • Constructed an app under client’s requirement, including front-end to back-end, aimed to help patients and doctors keep in touch with the progress in their recovery

  • Set up and managed the project using Git, implemented functions using related API, styled our app using Material UI

  • Managed the application state with Redux and React Hook, displayed multiple views in a single page using React Router

  • Categorized the recovery data and user profile on Json based on Firebase then visualized the data using Chart.js


Publication

Verification Code Recognition Using Convolutional Neural Network 11/2020

First Author of Abstract

  • Oral presentation on Conference on Soft Computing & Machine Intelligence (ISCMI 2020), Stockholm, Sweden during November 14-15, 2020

Honors and Awards

Scholarships & Academic Performance

  • Outstanding Graduate of Liaoning 2020

  • Model Student, NEU 2020

  • “Dimension Robotics” Scholarship for Outstanding Individual in Scientific Innovation, NEU 2020

  • “SMS Group” Top Grade Award for Excellent Students, NEU 2020

  • Second Class Scholarship, NEU 2020

  • Third Class “Dimension Robotics” Scholarship, NEU 2020

Competitions & Projects

  • National First Prize in the 15th National College Students Smart Car Competition 2020

  • Honorable Mention in MCM/ICM 2020

  • Second Prize in NEU, Liaoning Market Investigation and Analysis Contest 2020

  • Third Award, National Selection of CUMCM 2019

  • First Award, University Selection of CUMCM 2019

  • Third Prize in NEU, Internet+ Innovation and Entrepreneurship Competition 2019

Activities

  • Outstanding Individual in Summer Practice, NEU 2019

  • National Project, National College Students “One Belt One Road” Summer Practice 2019