I’m a CS student at Yuan Ze University. I’m interested in ML, backend, and more.

Currently working on Apache Yunikorn and constructing new Blog with better reading experience.

Projects & Experiences

RSS Friend - Blogroll for collecting friend’s blog post

• A few months ago, I came across a blog that featured a blogroll, a page that displays a friend's new posts on their blog without needing manual updates. Intuitively, I wanted to just copy the code, but the service being used was not accessible to me at that time. Since I had already familiarized myself with GitHub Actions and GitHub Pages, I came up with an idea to write a GitHub Pages version. Using GitHub Pages to deploy my API endpoints allows my Hexo markdown generator to use that API to generate a page for it. At that time, I had actually written a post about it: https://blog.ryankert.cc/2022/09/01/blogroll-setup/

Python Scraping & API - Yuan Ze University Course Data API

• Enabled data access through a REST API using the GET method for users.
• Developed a web scraper using BeautifulSoup in Python to extract course data from my school’s public website.
• Automated data updates through scheduled CI/CD pipelines using GitHub Actions and published the data via GitHub Pages.
  • Project’s GitHub

Segment Instances of Microvascular Structures

• Fine-tuned pre-trained models such as Yolov8, U-net, and MaskRCNN to enhance recognition accuracy.
• Implemented data augmentation and K-Fold cross-validation to prevent overfitting.
• Applied post-processing optimizations like mask dilation to improve precision. Monitored training outcomes with wandb to identify overfitting or underfitting scenarios.
• Google docs: https://docs.google.com/document/d/1GUXEYBuGit7vT9OZKeMnDz7c48aLupjpiYhyyPoSON0/edit?usp=sharing

Backend System Development - Crowd Management and Crisis Alert System

• Designed a system to send LINE messages based on user information stored in PostgreSQL with decision-making capabilities to prevent crowd overlaps and enhance crisis response.
• Leveraged Metis as the data-driven core to power our backend on Railway (deployed with Docker) utilizing RESTful APIs to communicate between services.
  • Project’s GitHub

Algorithm Experiences (mainly with C++)

• Implemented C++ STL (Standard Template Library) unordered set (hashing), set (red black tree), priority queue (heap), vector, string, deque with proper testings.
• LeetCode experiences - solved 400+ problems (80+ hard), Knight Badge (top 25% of contest ratings).
  • LeetCode Profile