IBM and Linux Foundation AI and Data (LFAI and Data) have joined together to create a “one-stop shop” for trusted data and AI artifacts in order to reduce duplication across teams when creating assets, as well as mitigate traceability, governance, risk management, lineage tracking, and metadata collection issues.
The announcement was made in a blog post by Animesh Singh, Christian Kadner, and Tommy Chaoping Li.
The Machine Learning eXchange (MLX), now in open source and open governance, is a single repository where all different asset types (e.g., datasets, models, and pipelines) are stored to be shared and reused across organizational boundaries, providing data scientists and developers with:
- Automated sample pipeline code generation to execute registered models, datasets, and notebooks
- Pipelines engine powered by Kubeflow Pipelines on Tekton, the core of Watson Studio Pipelines
- Registry for Kubeflow Pipeline Components
- Dataset management by Datashim
- Serving engine by KFServing
The Machine Learning Exchange also provides a marketplace and platform for data scientists to share, run, and collaborate on their assets. It can be used to host and collaborate on data and AI assets within a team and across teams.