Aichoo AIOS platform is primarily a data processing pipeline infrastructure with a kernel which orchestrates execution of Agents which are universal building blocks for any end-to-end Machine Learning pipeline.
Agents are also in principle similar to Linux/Unix tools which make these operating systems so powerful. Each agent is a self-contained class with clearly defined purpose of loading data columns from the AIOS Memory and creating new columns in return.
Agents classes are written primarily in Python but could as well be in R or any other language supported by Jupyter Notebooks as all execution happens on a Jupyter server. As long as Agents are developed to this simple specification, any Machine Learning pipeline can quickly be constructed from a library of such agents, hence simplifying model development and deployment of applications that rely on ML models.
We have been creating a public library of such agents, please see AIOS GitHub repository, and we would like to welcome all developers and data scientists to contribute to the AIOS_community repository.
Aichoo AIOS platform web service is now available online free for non-commercial use. Agents themselves are open-source.
Documentation how the platform works and how agents should be created can be found on the GitHub here.
If you have issues or questions, please add and discuss those also on the GitHub Issues page.