Machine Learning Real-Time

Wenkai Yang

929-***-**** # wenkaiya@buffalo.§ edu github. kevin123yang.com/Kevin123yang github.io ï linkedin.com/in/wenkai-yang Education

University at Buffalo Aug. 2021 – May 2025

Bachelor of Science in Computer Science, Buffalo, NY Core Coursework: Data Structures, Algorithms, Database Systems, Operating Systems, Software Engineering, Machine Learning

Projects

MMO Maze Game (Real-Time Backend Service) Jan 2025 – May 2025 Buffalo, NY

• Architected the backend for a real-time multiplayer maze game, supporting 20+ concurrent players with round-trip latency below 50ms via WebSocket (Flask-SocketIO).

• Developed a secure user authentication system using Flask-Login and session cookies, preventing unauthorized access and ensuring session integrity.

• Designed and optimized a MongoDB schema to persist player profiles, avatar data, and a live leaderboard, enabling seamless reconnect and state recovery.

• Integrated front-end JavaScript event handling to synchronize WASD movement and real-time maze rendering across clients.

• Deployed production-ready application using Render (PaaS) and MongoDB Atlas (DBaaS), maintaining stable online access during testing.

• Live Demo: mmo-maze-game.onrender.com — Source: GitHub Repo MNIST Classification & Support Vector Machine (SVM) Experiment May 2025 Buffalo, NY

• Implemented One-vs-All Logistic Regression and SVMs (linear & RBF kernels) for MNIST digit classification, achieving over 90% test accuracy.

• Tuned hyperparameters using scikit-learn and vectorized computations with NumPy, reducing training time by 40%.

• Analyzed kernel performance differences, demonstrating superior classification capability of the RBF kernel on non-linear data.

Image Processing: Panoramic Stitching Apr 2025

Buffalo, NY

• Developed a script to perform feature detection and homography estimation on images with varying perspectives using OpenCV and NumPy.

• Stitched 10+ sets of images into single panoramic photos, achieving alignment error below 1 pixel on test datasets.

• Visualized and documented results using Matplotlib to demonstrate algorithm effectiveness. Login with GitHub (OAuth 2.0 Integration) Feb 2025 Buffalo, NY

• Implemented a complete OAuth 2.0 flow using Python’s socketserver and requests, enabling GitHub login without registration.

• Integrated GitHub API to retrieve user data (username, email, profile picture) via three distinct endpoints.

• Secured API credentials with python-dotenv, adhering to best practices for token and secret management. Technical Skills

Programming Languages: Python, HTML, CSS, JavaScript Web Frameworks: Flask

Databases: MongoDB (Atlas)

ML Libraries: NumPy, scikit-learn, OpenCV

DevOps & Tools: Render, Docker, Git/GitHub, Visual Studio Code, PyCharm, Jupyter Notebook

Contact this candidate