Yifan Yang
NLP Algorithm Engineer at Tencent
Shenzhen, Guangdong Province, China
Summary
🤩
Rockstar🎓
Top SchoolYifan Yang is an NLP algorithm engineer with four years of experience applying spoken language processing and ASR techniques in industry and research, currently working at Tencent in Shenzhen. He blends academic rigor from MS work at NYU and ongoing PhD-level research with practical engineering—contributing to open-source ASR tooling like k2-fsa/icefall (adding Zipformer+CTC features and memory optimizations). His background spans end-to-end Q&A systems at Microsoft, medical NLP and product work at Tencent, and continual learning research, giving him a rare mix of model-level innovation and product-minded implementation. Colleagues rely on him to bridge decoding and training details into scalable systems, and he often surfaces practical tweaks (e.g., blank-skip and parameter tuning) that materially improve ASR performance.
5 years of coding experience
1 year of employment as a software developer
Summer Program, Summer Program, Summer Program, Summer Program at Georgia Institute of TechnologyMaster of Science - MS, Computer Engineering, Master of Science - MS, Computer Engineering at New York UniversityBachelor of Engineering - BE, Automation Engineer Technology/Technician, Bachelor of Engineering - BE, Automation Engineer Technology/Technician at Shanghai Jiao Tong University
Github Skills (10)
pytorch10
machine-learning10
speech-recognition10
automatic-speech-recognition10
python10
asr10
model-optimization9
nlp4
natural-language-processing4
tensorflow4
Programming languages (7)
Github contributions (5)
k2-fsa/icefall
Dec 2022 - Jan 2023
Role in this project:
ML Engineer
ML EngineerContributions:95 reviews, 9 commits, 167 PRs in 27 days
Contributions summary:Yifan primarily contributed to the project by implementing, modifying, and improving components related to automatic speech recognition (ASR) models and associated CTC decoding, especially regarding the Zipformer model. Their contributions included adding new functionalities, such as "Blank Skip" to the Zipformer+CTC model, and resolving code conflicts. They also optimized memory usage within the existing scaling functions and adjusted various parameters, demonstrating an understanding of model training and decoding processes.
yfyeung/yfyeung.github.io
Mar 2023 - Jul 2026
Contributions:270 pushes, 3 branches in 3 years 4 months
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.