Mohammad Zeineldeen
Speech Scientist at AppTek.ai
Aachen, North Rhine-Westphalia, Germany
Summary
👤
Senior🎓
Top SchoolMohammad Zeineldeen is a Speech Scientist and PhD candidate at RWTH Aachen University with a decade of engineering and research experience advancing end-to-end and streaming ASR systems. He combines academic rigor under Prof. Hermann Ney with industry impact at AppTek.ai, where he develops robust multilingual ASR and explores SpeechLLMs. His work spans acoustic and language modeling, neural LSTM variants, and practical deployable systems—evidenced by contributions improving LSTM layer normalization and stability in the widely used returnn framework. He has applied knowledge distillation techniques during a Google Speech research internship and routinely supervises student projects, bridging research and hands-on mentorship. Known for improving core RNN components and adding clear documentation, he brings both deep algorithmic insight and production-aware engineering to speech technology.
10 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at RWTH Aachen UniversityBachelor of Science - BS (With Distinction), Computer Science, 3.7/4.0, Bachelor of Science - BS (With Distinction), Computer Science, 3.7/4.0 at American University of Beirut
Arabic, English, German
Stackoverflow
312reputation
6kreached
9answers
1question
Github Skills (15)
normalize10
deeplearning-ai10
normalizing10
lstm10
recurrent-neural-networks10
deep-learning10
tensorflow10
python10
normalization10
backpropagation6
neural-network6
while-loop6
machine-learning6
do-while6
numpy6
Programming languages (2)
Github contributions (5)
rwth-i6/returnn
Oct 2018 - Mar 2022
The RWTH extensible training framework for universal recurrent neural networks
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 8 commits, 3 PRs in 3 years 4 months
Contributions summary:Mohammad primarily contributed to the `LayerNormVariantsLSTMCell` within the `returnn` repository, focusing on the recurrent neural network implementation. Their work involved fixing state representation, adding and managing the forget bias, and optimizing layer normalization, thus enhancing the stability and functionality of LSTM cells. These changes, along with the addition of documentation, show a commitment to improving and refining the core deep learning components of the project.
recurrent-neural-networksgputensorflowtheanodeep-learning
mmz33/N-Gram-Language-Model
Jan 2019 - Jan 2024
Contributions:38 pushes, 1 branch in 5 years 1 month
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.