The following page may contain information related to upcoming products, features and functionality. It is important to note that the information presented is for informational purposes only, so please do not rely on the information for purchasing or planning purposes. Just like with all projects, the items mentioned on the page are subject to change or delay, and the development, release, and timing of any products, features or functionality remain at the sole discretion of GitLab Inc.
Machine Learning and Artificial Intelligence promise smarter and powerful software and features. With the depth, breadth, and scale of the GitLab platform, there are many opportunities to leverage data science to make existing GitLab features more intelligent.
Watch a video overview of the GitLab Learn direction
This group is focused on how to extend GitLab functionality to provide additional value by leveraging ML/AI. This group will build on existing successful GitLab categories and features to make them smarter, easier to use, and more intelligent.
We are currently actively working on an ML model that automatically labels GitLab internal issues based on issue content. You'll see GitLab issues with the
automation:ml label that have been automatically labeled by our model. You can also provide training feedback to the model if it is incorrect by applying the
automation:ml wrong label. Internal employees can view a feed of these issues with probability data in Slack in the #feed-tanuki-stan channel.
We pursued this feature first as a way to get a data science workload working within GitLab's existing CI/CD as well as running on top of production GitLab data and interacting with the GitLab data model. This will set the foundation for work in our MLOps group and our other Applied ML categories listed above.
Signal detection is very hard in a noisy environment. GitLab intends to use ML to warn users of any signals that stand out against the background noise in several features:
Automatically categorizing and labelling is risky. Modern models tend to overfit, e.g. resulting in issues with too many labels. However, similar models can be used very well in combination with human interaction in the form of recommendation engines.
Because of their great ability to recognize patterns, neural networks are an excellent tool to help with scaling by anticipating needs. In GitLab, we can imagine:
Similar to DeepScan.
Identifying anomalous activity within audit events systems can be both challenging and valuable. This identification is difficult because audit events are raw, objective data points and must be interpreted against an organization's policies. Knowing about anomalous behavior is valuable because it can allow GitLab administrators and group owners to proactively manage undesirable events.
This is a difficult problem to solve, but can help to drastically reduce the overhead of managing risk within a GitLab environment.
Last Reviewed: 2021-02-14
Last Updated: 2021-02-14