About

avatar DID NOT LOAD

Yejin Cho

About Me

Interests

  • Internet Measurement & Protocol Analysis
  • IPv6 Dual-Stack Adoption
  • VPN Security Issues
  • Anycast & DNS Infrastructure
  • BGP Routing and Internet Outages
  • Graph Neural Networks for Network Modeling

Perspectives on Network & Internet Research

  • I want to understand why core protocols — like BGP and DNS — were designed under the constraints they were, and how those legacy decisions continue to shape the Internet's resilience and scalability today.
  • I find long-term, incremental protocol evolution fascinating: watching how anycast catchments shift or how congestion control algorithms like BBR mature reveals where theory meets operational reality.

Professional History

Before coming to the U.S., I worked as a software engineer in South Korea. I started at Mint Technology, an Apple-certified consulting company, where I contributed to the Hyundai ZET scooter rental platform and helped maintain and scale SSingSSing, Korea's leading scooter rental service with over 10k daily active rentals.

As AI models advanced rapidly, I realized that writing software alone would not be enough to keep up with where the field was heading. I became increasingly drawn to research — to understanding the systems and networks that underpin everything we build. That shift brought me to the U.S. to pursue a Master's in Computer Science.

Research before Internet Measurement

Before focusing on Internet measurement, I explored several other research areas during my undergraduate years -- mainly Sensor Networks, Recommendation System, LLM, and Computer Vision.

At the CAU Network Systems Lab (NSL), advised by Jeongyeup Paek, I studied Vehicular Ad-hoc Networks (VANET) and conducted a survey of broadcasting techniques, published at ICTC 2022. I also worked on deep-learning-based packet delay prediction, which led to a 5th-place finish in the ITU/Barcelona Neural Networking Center GNN Challenge 2023, where I built models trained on 500GB of PCAP data.

At the CAU Data Intelligence Lab, under Prof. Mucheol Kim, I contributed to an NSF-funded project on AI-based real-time optimal disaster evacuation route detection, developing the core logic for a Graph Neural Network and recommendation system-based algorithm. During this period, I also co-authored KRongBERT, a Korean-specific pretrained language model optimized for morphological segmentation, published in Information Processing & Management (2024).

I also gained hands-on experience in other areas through various projects: computer vision (image segmentation and object tracking), recommendation systems (content-based and collaborative filtering for a GitHub repository recommendation), and adversarial robustness evaluation (implementing FGSM, CW, PGD, and DeepFool attacks on PyTorch models for my capstone project, YourBench).

Publications

See my full list of publications on the Publications page.

Inspiration

전길남 (Kilnam Chon) — Connected Korea to the Internet (1982). 2012 Internet Hall of Fame inductee. Former ISOC (Internet Society) board member.

My dad — A filesystem researcher. I still remember peeking at his NTFS hex dumps when I was very young. He is the reason I first became interested in computer science.

Personal Life

I like making music in my free time! Check out my SoundCloud.