Developing machine learning models in a reproducable, comprehensible way is keen for AI teams, especially in the context of large enterprises where many Data Scientists work together on projects and where also common software development standards are key for delivering quality into production. In this post I'll explain by example what are simple but quite helpfull components for a more standardized way to develop machine learning models.
This is the personal blog of Michael Wellner. I'm an enthusiastic software engineer. I do and write about architecture, development and coaching.
When developing frontends I personally love to work with React and Redux due to its functional paradigms: You can easily develop, compose and test web components - Wondeful. But when developing larger projects one have to think about a good structure of components and code artifacts. When it comes to sharing components and (redux-)logic between several project a good composition is keen.
In the following I want to give you insights how I started structuring my projects almost a year ago from now and I still stick to this structure as I made good experience with it throughout the last year.