Kookmin University student wins 2nd place at IBM Bob Hackathon

Kim Chan-joong, a student majoring in artificial intelligence at Kookmin University / Courtesy of Kookmin University

A Kookmin University student won second place at the IBM Bob Hackathon, an international artificial intelligence (AI) competition held June 11, the university said Monday.

The award went to Kim Chan-joong, a student from the university’s department of artificial intelligence, who participated in the event individually.

The university said Kim was recognized for developing “Atlas,” a web application that visualizes the structure of a code repository as a city map.

The tool provides a single visual overview of a repository’s architecture, helping users quickly understand complex software projects where countless folders and files can make the overall structure difficult to navigate, especially for first-time users.

The competition was co-hosted by NativelyAI, the operator of the AI hackathon platform lablab.ai, and IBM. Participants developed software within a limited time frame using IBM’s AI development tool, “Bob.”

A total of 5,628 participants representing 1,672 teams from around the world took part in the competition, with 503 projects submitted.

The IBM Bob Hackathon results page shows that Kim Chan-joong, an artificial intelligence major at Kookmin University, won second place with his web application “Atlas.” Courtesy of Kookmin University

The university said that through Kim’s Atlas app, users can visualize the repository structure: Top-level folders are displayed as color-coded districts, while files are represented as buildings of different sizes based on their code volume.

Key files are highlighted as major landmarks on the map, allowing users to intuitively understand the overall structure and identify important files even in an unfamiliar code repository.

The map is automatically generated through repository structure analysis, with layout calculations handled on the server side. As a result, users can access the visualization directly through a web browser without requiring any separate installation.

The completed map supports zooming and navigation, displays file details when users hover over individual files, and allows the entire visualization to be downloaded as an image.

“I wanted to explore a fresh idea rather than pursue the common themes typically explored in AI projects. I had frequently found it time-consuming to understand the structure of code I encountered for the first time, and I wanted to solve that problem through an intuitive visualization approach like a city map," Kim said.

“I focused on keeping the scope manageable and building the core features first,” he added. “I hope to use this experience as a foundation to continue pursuing new ideas and experiments in the future.”