Junho Kim
Research Scientist at NAVER Corp
Seongnam-si, South Korea
Summary
🤩
Rockstar🎓
Top SchoolJunho Kim is a research scientist with nine years of experience specializing in generative models and computer vision, currently driving research at NAVER AI Lab. He has a strong practical background implementing and refining GAN and image-to-image translation architectures—contributions to TensorFlow projects include ResNet/DenseNet blocks, spectral normalization, custom losses, and official U-GAT-IT and StarGAN implementations. Junho’s work spans both industry R&D (NAVER, NCSOFT, NAVER WEBTOON) and medical/imbalanced-data research from earlier internships, giving him a breadth of applied ML experience. He holds bachelor’s and master's level training in computer science, mathematics, and NLP from Chung-Ang University, and actively recruits interns for generation-model research, reflecting a commitment to mentoring the next generation of researchers.
9 years of coding experience
2 years of employment as a software developer
Bachelor's degree, Computer Science and Engineering, Mathematics, Bachelor's degree, Computer Science and Engineering, Mathematics at 중앙대학교
Github Skills (20)
loss-functions10
python10
machine-learning10
image-translation10
generative-adversarial-network10
mask-rcnn10
keras10
deep-learning10
tensorflow10
neural-network10
faster-rcnn10
computer-vision10
image-processing9
batch-normalization9
mnist9
Programming languages (5)
Github contributions (5)
taki0112/Tensorflow-Cookbook
Feb 2019 - Feb 2020
Simple Tensorflow Cookbook for easy-to-use
Role in this project:
ML Engineer
ML EngineerContributions:110 commits, 1 PR, 107 pushes in 11 months
Contributions summary:Junho primarily contributed to the `ops.py` file, adding and modifying various functions related to neural network operations, loss functions, and normalization techniques within a TensorFlow environment. Their work includes implementing blocks like ResNet, DenseNet, and Squeeze-and-Excitation, alongside custom loss functions and pixel-shuffle upsampling, indicating a focus on designing and refining model architectures. The user also demonstrates proficiency in utilizing spectral normalization and defining specialized losses, enhancing the project's capabilities.
tensorflow-cookbookmlpcookbookimage-recognitionimage-classification
taki0112/UGATIT
Jul 2019 - Jan 2020
Official Tensorflow implementation of U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation (ICLR 2020)
Role in this project:ML Engineer
ML EngineerContributions:48 commits, 3 PRs, 47 pushes in 5 months
Contributions summary:Junho primarily contributes to the implementation of the U-GAT-IT model, a deep learning model for image-to-image translation. Their work involves writing code, modifying existing functions, and fixing parameters within the `UGATIT.py` and `ops.py` files. The user focuses on various aspects of the model, including the generator and discriminator architectures, indicating a strong understanding of the underlying model components and their configurations. These changes likely relate to improving model performance or addressing specific issues.
image-translationiclrcyclegandeep-learningadaptive
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.