Chanwoo Kim
Engineering Manager at Hyundai Motor Company (현대자동차)
Seoul, South Korea
Summary
👤
Senior🎓
Top SchoolChanwoo Kim is an engineering manager at Hyundai Motor Company with eight years of experience, bridging mechanical engineering with advanced ML development. He oversees air conditioning and cooling systems in Hyundai's Quality Division, while applying simulation skills in ANSYS and ABAQUS during design and testing. A research-oriented practitioner, his early work explored micro/nano-structures of vitreous carbon for water-oil separation and hot-stamping structural analysis. As an active ML engineer and open-source contributor, he has helped implement constrained beam search and M-CTC-T components in Hugging Face Transformers, and developed CentroidTripletLoss and VICRegLoss for metric learning. Based in Seoul, South Korea, he combines hands-on engineering leadership with a track record of impactful contributions to AI tooling and production-ready ML components.
8 years of coding experience
Baccalauréat, 기계공학부, Baccalauréat, 기계공학부 at 중앙대학교
English, Korean
Github Skills (17)
transformers10
pytorch10
language-model10
python10
metric-learning10
machine-learning10
language-models10
deep-learning10
natural-language-processing10
nlp10
testing9
speech-recognition9
computer-vision9
flax4
tensorflow24
Programming languages (7)
Github contributions (5)
KevinMusgrave/pytorch-metric-learning
Oct 2021 - May 2022
The easiest way to use deep metric learning in your application. Modular, flexible, and extensible. Written in PyTorch.
Role in this project:
ML Engineer
ML EngineerContributions:4 reviews, 19 commits, 13 PRs in 6 months
Contributions summary:Chanwoo primarily contributed to the implementation and testing of a new loss function, `CentroidTripletLoss`, within the `pytorch-metric-learning` framework. Their work included defining the loss function, writing the test code to validate this loss function, which involved debugging the code for return types. Additionally, the user implemented the `VICRegLoss`.
pytorchdeep-metric-learningdeep-learningcontrastive-learningcomputer-vision
huggingface/transformers
Feb 2022 - Jun 2022
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:ML Engineer
ML EngineerContributions:26 reviews, 5 commits, 10 PRs in 3 months
Contributions summary:Chanwoo primarily contributed to the implementation of constrained beam search functionality within the transformers library, a key aspect of natural language processing. They developed and refined classes for constrained beam search, including the `ConstrainedBeamSearchScorer` and `DisjunctiveConstraint` classes. They also addressed bug fixes, documentation, and test integration, ensuring the functionality integrated correctly with the existing codebase. The user's contributions included implementing the M-CTC-T model, including the feature extractor, processor, tokenizer and its tests.
pythonbertspeech-recognitionstate-of-the-artflax
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.