Jun Kim Joins Hugging Face to Support the MLX Community

Jun Kim Joins Hugging Face to Support the MLX Community

Hugging Face has welcomed Jun Kim, the creator and maintainer of oMLX, to its team. Kim will continue contributing to MLX and leading the development of oMLX.

MLX is Apple's framework for local artificial intelligence, with particular optimization for Apple Silicon. Hugging Face has supported the MLX ecosystem since receiving the framework as a Christmas present from Awni Hannun and Angelos Katharoulias in 2023. Its Hub also serves as a place where users can discover MLX models and contribute their own.

As the use of open, locally run AI grows, Hugging Face says it wants to support an ecosystem in which users can find tools suited to their needs.

What the Move Means for oMLX

The transition is expected to provide greater stability and potentially faster development for oMLX. Moving from a side project to a fully maintained and funded project will give Kim more capacity to guide contributors and plan for the long term.

The project will remain licensed under Apache 2.0, and Kim will continue leading it as before.

Supporting the Wider MLX Ecosystem

Hugging Face's broader goal is to help the community run local AI in different forms by providing the tools and building blocks required to do so. oMLX is expected to act as a testbed for new ideas while building on dependencies such as mlx-lm and mlx-vlm.

The organization also sees value in strong modeling and inference libraries and intends to contribute work upstream where appropriate. Hugging Face has collaborated with projects and companies including mlx-lm, mlx-vlm, and LM Studio. It hopes to strengthen cooperation with Cheng, Prince, Yagil, and their respective teams.

One specific focus is shortening the path from a Transformers model definition to a reference MLX implementation that can be used by multiple engines. This would allow each engine to concentrate on its own features while using a shared implementation. Because the Transformers library has become a reference point for machine learning model definitions, the aim is to make new Transformers models easier to run on MLX.