Jaewoo Song
Applied Scientist - Rufus at Amazon
Seattle, Washington, United States
Summary
👤
Senior🎓
Top SchoolJaewoo "Kyle" Song is an Applied Scientist at Amazon Rufus with eight years of experience building production-grade NLP and LLM systems, from pretraining Korean T5 models to deploying domain-adapted transformers for large-scale employee survey analysis. He has a strong track record at Amazon PXT Central Science where he engineered custom embeddings that outperformed Titan Embeddings, developed NER and topic-modeling pipelines in SageMaker, and designed benchmarks for model evaluation across varying context sizes. His research background from Penn includes studying LLMs as game masters and function-calling effects on gameplay, bridging academic insights with practical agentic AI applications. Fluent in both R&D and software engineering, he has accelerated training throughput, optimized long-context models, and shipped cloud-native features at AWS. Based in Seattle, he brings a rare combination of multilingual NLP expertise, hands-on systems optimization, and experience scaling models and tooling for real-world high-volume use cases.
8 years of coding experience
4 years of employment as a software developer
Exchange Program, Exchange Program at American UniversityBachelor of Science in Engineering, Computer Science and Engineering, Bachelor of Science in Engineering, Computer Science and Engineering at Yonsei UniversityMaster of Science in Engineering, Computer and Information Science, Master of Science in Engineering, Computer and Information Science at University of Pennsylvania
English, Korean
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (39)
pytorch10
python10
nlg10
machine-learning10
multimodal10
llm10
audio10
deep-learning10
natural-language-processing10
transformer10
speech-recognition10
nlp10
machine-learning-models9
transformers9
natural-language-generation9
Programming languages (5)
Github contributions (5)
devjwsong/gpt2-dialogue-generation-pytorch
Jul 2021 - Feb 2022
The PyTorch implementation of fine-tuning the GPT-2(Generative Pre-trained Transformer 2) for dialogue generation.
Contributions:27 commits, 3 PRs, 48 pushes in 7 months
fine-tuningpytorchgpt-2natural-language-processingnatural-language-generation
devjwsong/intent-capsnet-pytorch
Apr 2021 - Jun 2022
IntentCapsNet implementation in Pytorch
Contributions:6 commits, 1 PR, 3 pushes in 1 year 2 months
pytorchcapsule-networkintent-classificationnatural-language-processingnlp
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.