At Bilot, we keep focusing on making our business processes smoother, faster and more reliable. This includes everything from marketing side to technical, programming work. Diminishing risk of failures is even more important when having large amounts of code, as debugging and testing laborious. Coding the same things over and over is not only boring, but it is also prone to mistakes.
Having this in mind, we have been developing AI-JACK, our machine learning infrastructure, which significantly accelerates AI solution delivery and automation.
So what exactly is the AI-JACK?
Well, this tool is basically an open-source collection of code written in R, containing multiple modules and integrating all the necessary steps for delivering a machine learning solution.
This means that there are separate modules for data source connection, pre-processing, training and evaluating various types of models, versioning and eventually publishing.
The modelling is done using the H2O API, which is a state-of-the-art technology in opensource AI/ML development. As we aim to focus on applicable solutions, the modules are built in order to solve real-life business problems.
Most importantly, when conducting a project with AI-JACK, one only needs to modify configuration files, without the need to master R programming. All the needed parameters, models etc. are configured with specific files.
However, as problems can be way more complex, there is a possibility to create own models and solutions on top of AI-JACK. That’s why we decided that AI-JACK is open-source. We aim to allow everyone to use and enhance the solution as they wish.
Ok, but how do customers benefit from AI-JACK?
First and foremost, choosing AI-JACK speeds up an AI/ML project due to the fact that the code already exists. These solutions are consistent and ready-to-use. This means that some of the issues can be solved almost straight away only by configuring files. Even more complex problems can use the existing AI-JACK code during the process, though.
Another benefit is that as the code is open, users can customize and modify it in any way they want. Moreover, AI-JACK enables an easy deployment in multiple environment of customer’s decision, no matter if it’s a local PC, a virtual machine, server or a cloud platform.
We wrote the open source version of AI-JACK in R, but we are developing something similar in Python, which is likely to go open-source in the future.
Check out AI-JACK repo to know more or ask us directly at firstname.lastname@example.org!