Adel Moumen
AIML Research Intern at University of Cambridge
Cambridge, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolAdel Moumen is an AIML researcher and PhD student in Information Engineering at the University of Cambridge, with four years of hands-on experience building and optimizing speech processing systems. He is a core maintainer and back-end contributor to SpeechBrain, the popular PyTorch-based speech toolkit with 8,300+ GitHub stars, where he led ASR community efforts and cut issue backlog by half while driving major decoding and neural rescoring improvements that yielded a 38% WER gain. His work spans low-level CUDA/C++ accelerations (speeding Li-GRU training fivefold) to high-level recipe development (Whisper encoder-decoder on LibriSpeech achieving competitive WERs), reflecting a rare blend of research rigor and production-grade engineering. Now an AIML research intern at Apple, he combines academic publications and open-source stewardship with practical system fixes—like transducer loss device handling—that improve real-world model reliability. Known for organizing the SpeechBrain summit and integrating models into Hugging Face leaderboards, he excels at turning advanced speech research into widely usable tooling.
4 years of coding experience
2 years of employment as a software developer
Doctor of Philosophy - PhD, Information Engineering, Doctor of Philosophy - PhD, Information Engineering at University of CambridgeComputer Science, Computer Science at Réseau FIGUREErasmus + , Computer Science, Erasmus + , Computer Science at Swansea University / Prifysgol AbertaweMaster, Computer Science, Master, Computer Science at Avignon Université
Github Skills (4)
speech-processing10
cuda10
pytorch10
python9
Programming languages (7)
Github contributions (5)
speechbrain/speechbrain
Jul 2022 - Jan 2023
A PyTorch-based Speech Toolkit
Role in this project:
Back-end Developer
Back-end DeveloperContributions:459 reviews, 147 commits, 241 PRs in 5 months
Contributions summary:Adel's commits primarily focused on addressing issues related to device configurations within the transducer loss calculations. They modified the transducer loss implementation to ensure input tensors reside on the correct 'cuda' device. Additionally, the user made minor adjustments, such as correcting comments and fixing build settings in the project's documentation. These changes suggest a developer focused on improving a core speech processing module.
voice-recognitionasrspeech-recognitionspeech-separationspeaker-verification
Adel-Moumen/speechbrain
Jan 2022 - Mar 2025
A PyTorch-based Speech Toolkit
Contributions:2 PRs, 854 pushes, 94 branches in 3 years 2 months
pytorchspeech-toolkitspeech-recognitionspeech-synthesisspeech
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.