Kevin Chen

Data Science / Software Engineer

Hi! My name is Wei-Chieh (Kevin) Chen.

I'm a first-year master's student pursuing my master degree in Computational Data Science at Carnegie Mellon University. I'm especially passionate about several areas including data science / machine learning / deep learning / software development. I had worked at McKinsey & Company as a machine learning solution intern. My teammate and I built a product that accelerated the cost analysis of printed circuit boards. It’s now used by McKinsey’s global team and sold to other clients.

Before joining McKinsey, I worked as a digital consulting intern at Deloitte Digital. Helping companies to do the digital transformation.

Feel free to contact me! weichieh[at]andrew.cmu.edu

Education

Carnegie Mellon University

Master of Computational Data Science

GPA: 4.0/4.0

Finished coursework: Foundation of Computational Data Science

Ongoing coursework: Machine Learning, Deep Learning, Interactive Data Science, Data Science Seminar

September 2021 - December 2022

National Tsing Hua University

Bachelor of Science in Computer Science

GPA: 3.85/4.3

Honors: Academic Achievement Award * 2 (Students among the top 5%)

Selected coursework: Data Structures, Design and Analysis Algorithm, Computer Networks, Computer Architecture, Operating System, Massive Data Analysis, Cloud Programming, Software Project Management

September 2016 - June 2020

Skills

Programming Languages

Python, C, JavaScript, R, SQL, ASP.NET

Web Development

React, Flask, HTML, CSS

Frameworks

PyTorch, Numpy, Pandas, Scikit-learn, PySpark, Tensorflow

Tools

Azure, AWS, Git, Docker, JIRA, CircleCI, Firebase, Linux, FastAPI

Work Experience

McKinsey & Company

Machine Learning Solution Intern

• Developed the whole product that accelerated the cost analysis of printed circuit boards by React, Flask, and Docker with autonomous CI/CD workflow; Product was utilized by McKinsey global team and sold to other clients.

• Conducted research on object detection and trained an Efficient-det-d0 model with 94% accuracy rate. Increased 15% mean average precision and reduced 75% inference time compared to the original model.

• Examined an electronics company’s performance to develop competitive strategies by data analysis.

July 2020 - January 2021

Deloitte Digital

Digital Consulting Intern

• Assisted companies in executing digital transformation and identifying business weaknesses.

• Analyzed business sales process and explored opportunities for introducing digital tools to the analytical process.

• Visualized companies’ data on Salesforce to help C-suite track key performance indicators.

October 2019 - January 2020

Research

Adversarial Attack and Defense on Object Detection Model

Independent Study, National Tsing Hua University

• Implemented Carlini Wagner L2 Attack method to attack Single Shot Detection model based on machine learning. Achieved a success rate of 99% under white-box attacks.

• Utilized Adversarial Training and Clean Logit Pairing method to improve the robustness of model without affecting the accuracy rate. Improved robustness against adversarial attacks by 160.47% and 232.69%, respectively.

January 2019 - January 2020

Projects

• Sequence to Sequence Prediction | PyTorch · Deep Learning

Predicted phonemes contained in speech utterance based on CNN and LSTM; achieved 10.02 Levenshtein distance.

• Face Classification and Verification | PyTorch · Deep Learning

Classified 4000 people’s face based on customized ResNet model structure to achieve 85.5% accuracy rate.

Determined whether two random face image are the same person; achieved 99.9% AUC score

• Amazon Review Rating Prediction | NLP · PyTorch · Scikit-learn · Azure

Implemented word embedding on Amazon product review to train predictive models.

Built model by PyTorch and Scikit-learn, and deployed the model on Azure to create public endpoint to let other access the API of the model.

• Rookie Chef Helper | AWS · Python · Raspberry Pi · Web

Built by mobile application, AWS, web application, Raspberry Pi, Python

• Remote Caring System | AWS · Python · Raspberry Pi · Web

Led 3-person team to design a remote caring system website using AWS, Raspberry Pi, Python and web-related skills to track patients’ conditions without direct personal contact

An web application tracking quarantined patients' health conditions during COVID-19.

Awards

3 ^rd Place - Meichu Hackathon, October 2019