Sung-lin Yeh
PhD Candidate
United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolSung-lin Yeh is a PhD candidate in Computer Science at the University of Edinburgh with a decade of hands-on experience in speech and machine learning research. He has interned at Meta (FAIR and Voice Modelling/GenAI teams) and contributed to the SpeechBrain project at Mila, where he implemented multi-head location-aware attention and improved beam search in a widely used PyTorch speech toolkit. Skilled in back-end ML engineering, Sung-lin bridges rigorous academic research with production-oriented code, focusing on attention mechanisms that boost speech recognition performance. His trajectory from National Tsing Hua University to top-tier research labs highlights a knack for translating theoretical advances into practical toolkit improvements that benefit the broader open-source speech community.
10 years of coding experience
Doctor of Philosophy - PhD, Doctor of Philosophy - PhD at The University of EdinburghMaster's degree, Speech Signal Processing, Master's degree, Speech Signal Processing at National Tsing Hua University
Github Skills (9)
attention-mechanism10
pytorch10
machine-learning10
speech-recognition10
deep-learning10
python10
transformers9
beam-search9
audio-processing8
Programming languages (3)
Github contributions (5)
speechbrain/speechbrain
Sep 2020 - Sep 2021
A PyTorch-based Speech Toolkit
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:13 reviews, 320 commits, 3 PRs in 1 year
Contributions summary:Sung-lin primarily contributed to the implementation of multi-head location-aware attention mechanisms within the SpeechBrain toolkit, specifically focusing on the `nnet/attention.py` and `nnet/RNN.py` modules. Their work involved adding new attention modules, refining existing code, and fixing syntax errors. The contributions centered on improving the attention mechanisms to improve the performance of speech recognition models. Additionally, the user's changes extended to the beam search decoding process.
voice-recognitionasrspeech-recognitionspeech-separationspeaker-verification
30stomercury/30stomercury.github.io
Mar 2020 - Sep 2021
Github Pages template for academic personal websites, forked from mmistakes/minimal-mistakes
Contributions:145 pushes in 1 year 7 months
templategithub-pages-templatemmistakesmistakesjekyll
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.