Curriculum vitae

Hyeongyu Kim

Compiler Engineer, Compiler Team · Hyundai Motor Company

Academic CV (PDF) One-page résumé (PDF)

Experience

Hyundai Motor Company

Aug. 2026 – Present

Compiler Engineer, Compiler Team

  • Develop TVM-based compilation for custom-NPU deployment, including Relax/TIR lowering, tiling, scheduling, and memory mapping.
  • Work on hardware/software co-design to balance latency, bandwidth, throughput, and power.

Medical Artificial Intelligence Lab, Yonsei University

Mar. 2020 – Aug. 2026

Doctoral Researcher · Advisor: Prof. Dosik Hwang

  • Research on test-time adaptation, domain adaptation and generalization, and prediction under distribution shift.
  • Developed AcTTA, Buffer layers, and MIRA; co-developed CD-Buffer for adverse-weather object detection.
  • Worked on cross-modality 3D segmentation, multi-contrast MRI restoration, and diffusion-based image synthesis; member of the CrossMoDA 2021 winning team.

Education

Yonsei University

Mar. 2020 – Aug. 2026

Ph.D. in Electrical and Electronic Engineering

Advisor: Prof. Dosik Hwang. Degree awarded August 28, 2026.

Yonsei University

Mar. 2014 – Feb. 2020

B.S. in Electrical and Electronic Engineering

Research interests

Test-time learning
Test-time adaptation, calibration, unsupervised domain adaptation, domain generalization

Medical imaging
3D segmentation, multi-contrast MRI restoration and synthesis, diffusion models, flow matching

Reliable deployment
Prediction under distribution shift, efficient inference, hardware-aware model deployment

Selected publications

Full publication list
  1. NeurIPS 2026 Accepted

    MIRA: Mutual Information guided calibration for Reliable Test-Time Adaptation

    Hyeongyu Kim* , Mingyeong Jang* , Youngjun Song , GeonHui Han , Dosik Hwang

    * Equal contribution

    Paper
  2. CVPR 2026

    AcTTA: Rethinking Test-Time Adaptation via Dynamic Activation

    Hyeongyu Kim* , Geonhui Han* , Dosik Hwang

    * Equal contribution

    Paper Project
  3. NeurIPS 2025

    Buffer layers for Test-Time Adaptation

    Hyeongyu Kim* , Geonhui Han* , Dosik Hwang

    * Equal contribution

    Paper Code

Industry & clinical projects

Fetal Ultrasound Image Enhancement

2025

Samsung Medison / Severance Hospital

Developed diffusion- and flow-matching-based enhancement for denoising, super-resolution, and artifact suppression without high-resolution reference data.

MRI-based Alzheimer’s Disease Diagnosis

2023

Samsung Electronics

Developed a cyclic inference-based diagnostic algorithm using brain MRI; associated patent application DP-2023-0002.

Pediatric Chest X-ray Dataset & Disease Detection

2023

Pediatric medical imaging data project

Constructed a pediatric chest X-ray dataset and trained models for lung disease detection.

Automated Spinal Analysis

2021

Rayence

Developed vertebral segmentation and spinal curvature/Cobb angle analysis from X-ray imaging.

Honors & scholarships

  • 2026
    Outstanding Reviewer
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2026)
  • 2024
    4th Rank · Aorta Segmentation Challenge
    MICCAI 2024 Challenge
  • 2024
    Research Scholarship
    Hyundai Motor Group
  • 2021
    1st Rank · CrossMoDA 2021
    MICCAI Cross-Modality Domain Adaptation Challenge
  • 2021
    Excellence Award · AI-based Image Diagnosis Hackathon
    Ministry of Trade, Industry and Energy
  • 2021
    Excellence Award · Medical Data AI Datathon
    Ministry of Science and ICT / National Information Society Agency
  • 2021
    International Collaborative Research Scholarship
    Yonsei University
  • 2020
    Fellow Scholarship
    Yonsei University

Patent applications

  • 2025
    Artificial neural network apparatus and test-time adaptation method for inference
    DP-2025-0773 · Application, South Korea
  • 2025
    Isotropic MRI volume reconstruction apparatus and method
    DP-2024-1961 · Application, South Korea
  • 2024
    Apparatus and method for generating multi-contrast magnetic resonance images
    DP-2024-0932 · Application, South Korea
  • 2023
    Dementia diagnosis system and control method based on cyclic inference-driven artificial intelligence
    DP-2023-0002 · Application, South Korea
  • 2023
    Domain-generalized segmentation model training method
    DP-2023-1013 · Application, South Korea

Technical skills

Programming
Python, C/C++, MATLAB
ML frameworks
PyTorch, TensorFlow, CUDA, OpenCV
Compilers & tools
TVM (Relax, TIR), Git, Docker, Linux
Languages
Korean, English

Teaching

Introduction to Bioengineering for Electrical and Electronic Engineering

Mar. 2020 – Jun. 2020

Teaching Assistant · Yonsei University

Reviewer service

Conferences
CVPR 2026, 2025 · Outstanding Reviewer (2026)
NeurIPS 2026
MICCAI 2025
ICPR 2024
Journals
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)
Medical Image Analysis
Biomedical Engineering Letters (BMEL)