O’Reilly’s Ben Lorica recently published a blog post entitled “6 Reasons Why I Like KeystoneML“.    In the article he states “Since its release in May, I’ve had a chance to play around with KeystoneML and while it’s quite new, there are several things I already like about it” and goes on to discuss those things.   Without giving too much away, these are well-aligned with many of the key objectives we had when we set out to build a new pipelines layer and API for BDAS and include:  type safety for programability and debugging, support for the full machine learning lifecycle, extensibility to new data sources, scalability using Apache Spark, hooks for automatic optimization of methods, physical resources and hyperparameter tuning (thanks to TuPAQ).

Ben L. was inspired to write the post based on his discussion and interview with AMPLab faculty member Ben Recht.   The full interview can be heard here and can also be found in the post.

Ben L. has a knack for discovering important technologies early (he was one of the early industry people to get behind Spark) so this is a real vote of confidence for the KeystoneML technology, which is still in an Alpha release.