Sang-gil Lee
Research Scientist at NVIDIA
United States, Republic Of Korea
Summary
🤩
RockstarSang-gil Lee is a research scientist with a decade of experience specializing in deep generative models for sequences, particularly speech and audio. Based at NVIDIA, he combines research rigor with hands-on ML engineering, contributing to notable open-source projects like Microsoft’s NeuralSpeech where he implemented and migrated PriorGrad acoustic and vocoder modules. His work spans model integration, fast sampling optimizations, and practical fixes that improve deployment compatibility and inference for text-to-speech pipelines. Fluent in both research and production concerns, he brings a pragmatic focus on reproducibility and performance to complex audio modeling problems. An interesting facet of his profile is the blend of academic-style research depth with direct contributions that make large speech models more usable in real-world systems.
10 years of coding experience
Github Skills (11)
diffusion-models10
pytorch10
machine-learning10
speech-synthesis10
deep-learning10
modeling9
python9
trainings9
continuous-deployment8
ml-deployment8
tensorflow4
Programming languages (2)
Github contributions (5)
microsoft/NeuralSpeech
Mar 2022 - Mar 2022
Role in this project:
ML Engineer
ML EngineerContributions:31 commits, 1 PR, 12 pushes in 24 days
Contributions summary:Sang-gil primarily contributed to the PriorGrad-acoustic and PriorGrad-vocoder projects within the repository, focusing on implementing and migrating various acoustic model components and related utilities. Their work includes migrating PriorGrad-acoustic and vocoder modules, integrating text inference support, and fixing potential compatibility issues. They also modified the code to support fast sampling iterations and correct mel shape issues.
NVIDIA/BigVGAN
Jun 2022 - Sep 2024
Official PyTorch implementation of BigVGAN (ICLR 2023)
Contributions:7 releases, 2 reviews, 14 PRs in 2 years 3 months
audio-generationaudio-synthesismusic-synthesisneural-vocodersinging-voice-synthesis
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.