Yoach Lacombe
Machine Learning Engineer at Cantina Labs
France
Summary
🤩
Rockstar🎓
Top SchoolYoach Lacombe is a machine learning engineer with six years of experience building state-of-the-art audio and generative models, currently driving ML work at Cantina Labs after a tenure at Hugging Face. He combines a rigorous applied math and AI background from Télécom Paris and ENS Paris-Saclay with practical production integrations—most notably contributing Bark and Stable Speech features to the widely used Hugging Face Transformers and Parler-TTS projects. His work spans model implementation, efficient inference (offload support) and novel prompt-based conditioning mechanisms, showing strength at the intersection of research and engineering. Fluent in deploying research-quality speech models into libraries relied on by the community, he brings both academic depth and hands-on open-source impact.
6 years of coding experience
3 years of employment as a software developer
MRes MVA - Mathematics, Vision, Learning, Applied Mathematics & AI, MRes MVA - Mathematics, Vision, Learning, Applied Mathematics & AI at École normale supérieure Paris-SaclayClasses préparatoires aux grandes écoles (CPGE), Mathematics and Physics (MPSI/MP*), Classes préparatoires aux grandes écoles (CPGE), Mathematics and Physics (MPSI/MP*) at Lycée KléberDiplôme d'ingénieur, Applied Mathematics and AI, GPA: 4.0/4.0, Diplôme d'ingénieur, Applied Mathematics and AI, GPA: 4.0/4.0 at Télécom ParisMaster in Management, Master in Management at HEC Paris
Github Skills (18)
pytorch10
integrate10
python10
model-driven10
attention-mechanism10
machine-learning10
integrations10
transformer-models10
model-building10
deep-learning10
natural-language-processing10
data-integration10
modeling10
model-driven-development10
transformers9
Programming languages (6)
Github contributions (5)
huggingface/parler-tts
Apr 2024 - Jan 2025
Inference and training library for high-quality TTS models.
Role in this project:
ML Engineer
ML EngineerContributions:15 reviews, 53 PRs, 30 pushes in 9 months
Contributions summary:Yoach's commits focus on adding new features for Stable Speech models, specifically integrating prompt concatenation into the modeling code. They modified the decoder and other model components to accommodate prompt hidden states and attention masks, suggesting the development of a new prompt-based mechanism. The changes include adapting the model for causal language modeling and extending the conditional generation capabilities.
huggingface/transformers
Jun 2023 - Dec 2024
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:ML Engineer
ML EngineerContributions:504 reviews, 116 PRs, 50 pushes in 1 year 6 months
Contributions summary:Yoach made significant contributions to integrating and implementing the Bark speech generation model within the Hugging Face Transformers library. Their work included adding the initial implementation of the Bark model, incorporating a conversion script, and supporting end-to-end inference functionality. Furthermore, they added functionalities such as offload support, improving the model's efficiency. The user's contributions demonstrate a focus on audio generation and machine learning model integration.
pythonbertspeech-recognitionstate-of-the-artflax
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.