During the past few weeks, the endtoend.ai website went through major changes in both the style and the content. In this post, we want to share the details and the rationales behind the changes.
The endtoend.ai website is created using Jekyll and is hosted on GitHub Pages. The previous theme was inspired by the Google Developers website. This website incorporated Material Design that offered intuitive yet pleasing pages. Specifically, we used Material Design Lite, a front-end template that simplifies the process of using Material Design, using the jekyll-mdl theme.
We found over the years that Material Design Lite was well-suited for creating beautiful front pages with card-based layouts, but came short when the main content of the page was text. To improve readability, we have experimented with various styles and found some success, but the website soon became very tedious to manage.
As a result, we started looking for alternative styles that better fit our use case. For machine learning bloggers, Medium has been a popular choice, as it offers an excellent reading experience.
Fortunately, while searching for alternatives, we found a mediumish: a Jekyll theme that emulated Medium’s style. With this theme, we could enjoy the visual attractiveness of Medium-style posts while not being hindered by Medium’s restrictions. Furthermore, the transition plan from the Material Design Lite theme was straightforward, since both were Jekyll themes.
Upcoming Content: Tutorials
Content-wise, we also have exciting things planned. Previously, blog posts on endtoend.ai have been largely focused on technologies directly related to machine learning. These long posts were written to summarize or clarify complex machine learning concepts and were therefore not geared towards beginners.
Recently, we began writing more beginner-friendly blog posts to share information that may benefit those that are new to the field. However, this resulted in a confusing situation, since the introductory posts and posts that assumed prior knowledge were listed in the same place.
To solve this problem, we will now have a Tutorials category for posts. Posts with this category will be written assuming minimal prior knowledge and will be gathered in a separate page. On the front page, the usual posts that go deeper into detail will be shown. By creating this separation, we hope that it will minimize the reader’s confusion.
Finally, we plan to write more tutorial posts on topics that are not directly related to machine learning. Machine learning benefits from a solid background in science and engineering. These upcoming posts on writing clean code and managing experiments will help those interested in machine learning just as much as other more machine-learning related posts.