SNC Project

• Duration: Ongoing

• Role: Original Member, Team Leader

• Description:
  Collaboration between Quantum Computing (Brian) and Astrophysics (Jiwoong) to explore Artificial Intelligence and interdisciplinary research.
  Focused on understanding and explaining concepts using Python and C++.

• Link

• Thanks!

  1. Sunje Keum

  2. Kim Hangyeol Park

Hotel Website Development 

• Duration: November 12, 2024 - Ongoing

• Goal
  This project aimed to create a fully functional hotel management system to simplify operations for a hotel I personally own. 

• Role
  I developed both the frontend and backend components of the website entirely from scratch. The platform allows for seamless room management, reservation handling, and user interaction, and is currently being used in a real-world setting. 

• Tech Stack

Django, HTML, CSS, JavaScript, MySQL

• Challenges
  As it was my first web development project, I faced challenges throughout the entire stack—from setting up the backend framework to configuring the deployment environment. Managing the server was particularly difficult. 

• Solution
  I overcame most challenges through self-guided learning and online documentation, particularly in setting up and securing the server for stable operation. 

• Result
  The website was successfully deployed and is currently in use for day-to-day hotel operations. 

• Reflection
  This project gave me practical knowledge of Django and taught me how to manage web servers. It also provided firsthand experience in building and maintaining a complete web application in a real-world context. 

• git Link

• Hotel Link

Faulty Text AS Center

• Duration: January 12, 2025 - March 01, 2025

• Goal
  This project aimed to restore obfuscated Korean text—particularly texts intentionally written in distorted or irregular formats, such as product reviews designed to evade automated translation or filtering systems. 

• Role
  I was primarily responsible for designing and implementing the core programs, including developing Transformer-based models tailored for the text restoration task. 

• Tech Stack
  Python, Transformer-based deep learning models 

• Challenges
  The most challenging part was data classification. It was difficult to create a consistent dataset that accurately represented distorted versus clean Korean text. 

• Solution
  I addressed this issue by referencing several academic papers and applying research-driven preprocessing techniques to improve data quality and class separation. 

• Result
  While the model's overall performance and accuracy were somewhat limited, the project provided valuable insights into handling complex, real-world NLP challenges. 

• Reflection
  Through this experience, I gained a deeper understanding of the limitations of natural language processing in noisy or adversarial text environments. It also strengthened my practical skills in model implementation and research-based problem solving. 

• Link

• Thanks!

  • Younghoon Lee

  • Eunseok Lee

  • Jiwon Lee

SpectraTrail

• Duration: January 10, 2025 - Ongoing

• Goal
  SpectraTrail is an ongoing project that aims to identify stars and extract their physical properties—such as name, type, and spectral data—by analyzing images of the night sky captured via smartphone. 

• Role 

I serve as the team leader, overseeing the entire project workflow including AI model design, mobile application planning, and web development architecture. 

• Tech Stack
  Python, PyTorch, OpenCV 

• Challenges
  As the project is currently in its early planning and prototyping stage, major technical challenges have yet to arise. However, we anticipate future difficulties in star detection under noisy conditions and differentiating overlapping spectral lines. 

• Current Progress
  The core concept, system structure, and technology roadmap have been established. Implementation will proceed iteratively as development time becomes available. 

• Reflection
  Leading SpectraTrail has sharpened my project planning and technical coordination skills. It also motivates me to bridge AI, astronomy, and mobile computing into a unified system capable of democratizing astronomical observations. 

• Thanks!

  • Jiwoong Choi

• Link

SpectraNeura

• Duration: January 10, 2025 - April 20, 2025

• Goal
  SpectraNeura was a collaborative initiative aimed at exploring various deep learning approaches to identify the most effective models across diverse tasks. The goal was to apply, compare, and enhance deep learning techniques through hands-on experimentation. 

• Role
  I focused on implementing models, reading and presenting research papers, and incorporating novel ideas to improve baseline performance. My role also involved translating theoretical insights into practical experiments. 

• Tech Stack
  PyTorch

• Challenges
  One of the main challenges was identifying the optimal model architecture for each unique task. It required significant tuning and understanding of underlying data distributions. 

• Solution
  I consistently explored recent papers and adapted their techniques to our experimental framework, aiming to achieve measurable improvements over baselines. 

• Result
  In some tasks, the models I developed achieved the highest accuracy among the team, while in others they fell short—providing valuable insights into the trade-offs of different approaches. 

• Reflection
  This project helped me realize that deep learning is not always superior to traditional machine learning methods. I learned to approach problems with simplicity first, and that understanding the fundamentals often leads to better results than overly complex designs. 

• Thanks!

  • Jeongnyeon Kim

  • Jihyeon Yoon

  • Junhong Min

• Link