Minchan Kim

profile_image.png

“I’m looking for Ph.D. positions for Fall 2026.”

Hi there, I’m Minchan 👋

I’m an undergraduate at Seoul National University, majoring in Computer Science. I’m currently working in SNU Vision & Learning Lab, advised by Prof. Gunhee Kim. I’ve also had the opportunity to collaborate on research projects with Prof. Yejin Choi at Stanford and Prof. Niloofar Mireshghallah, Prof. Akari Asai, and Prof. Graham Neubig at CMU.

My research focus lies in the intersection of machine learning, databases, and cryptography: machine learning (privacy-preserving machine learning, long-term memory systems, generalist imitation agents), databases (vector databases, approximate nearest neighbor, multimodal data compression), cryptography (homomorphic encryption, multi-party computation, private information retrieval)

My vision is to develop a secure long-term memory system inspired by the human hippocampus—one that continuously collects, stores, and utilizes an individual’s lifetime of action data. Through this work, I aim to create a generalist imitation agent that shares your memories and acts on your intent. To achieve this, the system must (i) seamlessly capture everything you see and hear, (ii) securely store your memories, and (iii) think, speak, and act like you.

If you share these interests or would like to explore potential collaborations, please feel free to reach out!

News

Jun 30, 2025 Our paper PPMI got accepted to Crypto 2025 Workshop! :smile:
Jan 28, 2025 Our paper CANVAS got accepted to ICRA 2025! :smile:
Dec 15, 2024 Our paper CANVAS won the Outstanding Paper Award at NeurIPS 2024 Workshop! :smile:
Sep 30, 2024 Our paper CUPID got accepted to ICDM 2024 Workshop! :smile:
Jul 01, 2024 Our paper ESREAL got accepted to ECCV 2024! :smile:

Publications

† : equal contribution
  1. d2e_preview.png
    D2E: Scaling Vision-Action Pretraining on Desktop Data for Transfer to Embodied AI
    Suhwan Choi, Jaeyoon Jung, Haebin Seong, and 7 more authors
    Under Review
  2. mipbench_preview.png
    MIP-Bench: Can LLMs Implicitly Personalize Answers for Different Users by Implicitly Leveraging Long-Term Memory?
    Minyeong KimMinchan Kim, Seungone Kim, and 6 more authors
    Under Review
  3. ppmi_preview.png
    PPMI: Privacy-Preserving LLM Interaction with Socratic Chain-of-Thought Reasoning and Homomorphically Encrypted Vector Databases
    Yubeen BaeMinchan Kim, Jaejin Lee, and 4 more authors
    Under Review, Crypto 2025 Workshop
  4. canvas_preview2.png
    CANVAS: Commonsense-Aware Navigation System for Intuitive Human-Robot Interaction
    Suhwan Choi, Yongjun ChoMinchan Kim, and 9 more authors
    ICRA 2025, NeurIPS 2024 Workshop (Outstanding Paper Award)
  5. esreal_preview2.png
    ESREAL: Exploiting Semantic Reconstruction to Mitigate Hallucinations in Vision-Language Models
    Minchan Kim, Minyeong Kim, Junik Bae, and 3 more authors
    ECCV 2024
  6. cupid_preview3.png
    CUPID: A Real-Time Session-Based Reciprocal Recommendation System for a One-on-One Social Discovery Platform
    Beomsu Kim, Sangbum KimMinchan Kim, and 7 more authors
    ICDM 2024 Workshop