Kamil Akesbi
Machine Learning Engineer at Deezer
Paris, Ile-de-France
Summary
🤩
Rockstar🎓
Top SchoolKamil Akesbi is a Machine Learning Engineer based in Paris with five years of experience building and researching deep learning systems for audio and vision. Trained at Centrale Lille and ENS Paris-Saclay (MVA), he has blended academic rigor with product-focused work across Deezer, Hugging Face and industry internships, tackling audio denoising, fingerprinting, music generation and medical imaging. He is an active contributor to Hugging Face Transformers, adding Whisper timestamping, short-form audio support and integrating the Descript-Audio-Codec—work that materially improves speech and audio model tooling for the community. Comfortable moving models from research to production, he combines strong math foundations with practical engineering (Python, Docker, cloud) and experience building annotation pipelines and interpretability tools. Notably, he pairs music-theory-informed analysis of transformer attention with hands-on audio model engineering, a rare mix that helps bridge signal insight and scalable ML systems.
5 years of coding experience
3 years of employment as a software developer
Master's degree MVA - Mathématiques Vision Apprentissage, Master's degree MVA - Mathématiques Vision Apprentissage at ENS Paris-SaclayMaster 2 (M2) Mathématiques, Master 2 (M2) Mathématiques at University of Lille 1 Sciences and TechnologyApplied mathematics, Applied mathematics at Universidad de ChilePCSI/PC**, PCSI/PC** at CPGE lycee Janson de Sailly, ParisBaccalaureat scientifique, Baccalaureat scientifique at Lycee Descartes, RabatMaster's degree Engineering, Master's degree Engineering at Centrale Lille
French, Arabic, English, Spanish
Github Skills (19)
pytorch10
integrate10
python10
model-driven10
whisper10
machine-learning10
audio-processing10
integrations10
model-building10
deep-learning10
transformer10
speech-recognition10
nlp10
data-integration10
modeling10
Programming languages (2)
Github contributions (5)
huggingface/transformers
Apr 2024 - Aug 2024
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:
ML Engineer
ML EngineerContributions:68 reviews, 20 PRs, 4 pushes in 4 months
Contributions summary:Kamil primarily contributed to the `huggingface/transformers` repository by implementing and integrating features related to audio models, specifically focusing on the Whisper architecture. Their work included adding support for token-level and word-level timestamps within the Whisper pipeline, as well as incorporating functionality for short-form audio processing. The user also developed and integrated the new Descript-Audio-Codec (DAC) model within the Transformers library, demonstrating their expertise in audio model development. These contributions enhance the capabilities of the Transformers library for speech-related tasks.
pythonbertspeech-recognitionstate-of-the-artflax
kamilakesbi/transformers
Apr 2024 - Aug 2024
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Contributions:262 pushes, 11 branches in 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.