Join

Graduate Students

For All Students:

As a faculty member in GIST AI, I have a few slots available to host PhD students each year. Before reaching out, please check whether your research interests align with our recent works (publications after 2024)

• I generally accept PhD students who have either conducted research with me before or have at least one first-authored publication in a venue relevant to our field. (e.g., NeurIPS, AAAI, ICLR, ICML, KDD, etc.)
• I also accept MS students who would like to continue their academic career. MS students in our lab are expected to publish (or have accepted) at least one first-authored paper in one of the above conferences before their thesis defense.

Please check the links below periodically for information such as official application periods and required documents:

• Call For International MS/PhD Applicants: [Link] [Apply]
• Call For Korean MS/PhD Applicants: [Link]
• GIFT Program for GIST Undergrads [Link]

Due to the large number of requests, I may not be able to respond to all emails. After submitting your official application, please send your CV through this form so that I can track your interest. Please note that I only write recommendation letters for students I have worked with, so be sure to obtain recommendations from your current mentors.

Undergraduate Students

Undergraduate students may join our lab through Undergraduate Thesis I/II (3 credits each) or apply for G-SURF. As an undergraduate students in our lab, you will participate in research discussions, build prototypes, host events, and write documents to help scale our projects and team efforts.

Required Qualifications

• Self-motivated individuals who enjoy tackling unsolved challenges
• Solid foundations in mathematics and computer science
• Strong technical, analytical, and communication skills
• Commit 16+ hours/week when you are with us

Preferred Qualifications

• Avid reader (e.g., enjoys books like Gödel, Escher, Bach)
• Passionate writer (e.g., blogger or book author)
• Machine learner (e.g., builds off-the-wall ML projects, contributes to open source)
• Full-stack developer (e.g., turns side projects into production-ready systems)
• Medal chaser (e.g., enjoys winning in Kaggle, ICPC, or Codeforces)
• Skilled puzzle solver or competitive gamer (e.g., high elo in LoL, Chess, Go)

General tips

• Take as many math courses as well as CS/EE/AI courses as possible; this will enable you to investigate a wider range of topics. Don’t be afraid to implement your ideas. If you are interested in our research, familiarizing yourself with this Kaggle competition would be helpful.

Students outside Korea

• Thank you for your interest in our lab. GIST offers an exchange program for regular semesters and a Global Internship Program (GIP) for summer vacations. You can apply for these programs by including my name in your application. The GIP program facilitates an 8-week visit to GIST over the summer, with the application period beginning in February. For more information, please visit this website: https://ipa.gist.ac.kr/. (Due to the large number of requests, I cannot write a letter of recommendation for GIP program.)

Lab Interview Process

In addition to the official department admission procedures, we conduct our own lab-specific interview process. This helps us better understand potential lab members and ensure a good fit. Our process consists of a one-hour interview with the applicant and current lab members.

During this interview, I generally ask the following questions:

• How would you describe yourself?
• Can you tell us about your project/research experience?
• What potential research ideas would you like to explore? (It would be great if you could align your ideas with our previous works)
• How do you think you can contribute to our lab?
• What are your career plans after completing your MS/PhD degree?

Previous Projects

Listed below are some projects I worked with (undergraduate) students. See my publication for a complete list.

At GIST

• Leveraging GIF to elevate AI reasoning [arXiv]
• Skill Reinforcement Learning [arXiv]
• Unsupervised Environment Design [arXiv]
• Misalignment analysis [KDD-25]
• Diffusion-based Reinforcement Learning on ARC-AGI [arXiv]
• Evaluating reasoning capabilities of LLMs [ACM-TIST-25]
• Multiple-choice analogy benchmark [EMNLP-24]
• Gym environment for RL training on ARC-AGI [CoLLAs-24]
• Footage from IJCAI-24 [IJCAI-24]
• Game with Abstraction and Reasoning Corpus [IJCAI-24]
• Seven papers at Korea Software Congress 2023 [KSC-23]
• Mimicking Human Solutions with Object-Centric Decision Transformer [ICMLW-23]
• Abstraction and Reasoning Challenge with Decision Transformer [KCC-23]
• Object Detection on ARC Problem Using Graph Abstraction [KCC-23]
• Engineering Prompts for the ARC Challenge [KCC-23]

Before GIST

• Playgrounds for Abstraction and Reasoning [NeurIPSW’22] [KSC’22] [Poster] [Happy ARC Day]
• 2022 Summer Intern’s talk [Subin-5min] [Prin-5min] [Shiekh-5min]
• Classification of Goods Using Text Descriptions With Sentences Retrieval [TIST]
• Customs Fraud Detection in the Presence of Concept Drift [ICDMW’21] [Video]
• Response to COVID-19 with Probabilistic Programming [Frontiers in Public Health’22] [GitHub]
• Hyperbolic Interaction Model for Category Classification in E-Commerce [KSC’21] [Video]
• Active Learning for Custom Fraud Detection [TKDE’22]
• Hierarchical POI Embedding with Heterogeneous Information [Slides] [GitHub]
• Customer Revisit Prediction using Macroscale Mobility [KSC’19] [URP Report]