CLAIMED is a library of re-usable coarse-grained data processing components to create Data & AI pipelines without programming skills
Lead Data Scientists and Domain Experts contribute to the library to create opinionated, tested and re-usable components which are consumed by citizen data scientists and developers which enables them to create state of the art Data & AI workflows
An open source project is only as strong as its community. IBM wants to grow the community around CLAIMED and Elyra since both projects are very strategic open source projects for IBM and RedHat
The initial repository was created in 2015 and was originally used to support the online courses IBM provides on Coursera.org and EDX.org. From the very positive feedback IBM received from the learners, we decided to create a general purpose library for AI, Machine Learning, ETL and Data Science.
No. Open Source and open standards are the key principles of CLAIMED. Therefore CLAIMED can be used in various contexts and therefore doesn't compete but integrate.
We are actively looking for developers and data scientists to use the library for the daily work including production ready software. We also want them to report issues, fix issues via pull request, participate in our discussions and contribute new components to the library
CLAIMED can be consumed in many ways and we encourage and support all scenarios. As each CLAIMED component is backed by a jupyter notebook or (R|python|bash) script and defines a clear interface they can be invoked directly from source code or from a command line. This way a Data & AI pipeline can be build by arbitrary code or shell scripts. The next level is using docker. As each CLAIMED component is automatically compiled into a container image, a set of "docker run" commands will do the job. Finally, CLAIMED also creates Kubeflow Pipeline Component specifications automatically, therefore, CLAIMED can be used in any Kubeflow Pipeline setting, where the Gold standard is using ML Exchange as component repository and Elyra as graphical pipeline editor.