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.
Today, data scientists piece together data, tooling, and frameworks to get bespoke data science workloads running and producing business value. Data scientists then hand off models to engineering teams to attempt to deploy them to production. These teams speak different languages, use different tools, and have completely different workflows. This makes it very difficult to deploy data science workloads to production, increasing time to value and costs. We want to make it faster and easier for organizations to get data science efforts to production as efficiently and collaboratively as possible.
Watch a video overview of the GitLab ModelOps direction
This group will be focused on enabling data teams to build, test, and deploy their machine learning models. This will be net new functionality within GitLab and will bridge the gap between DataOps teams, data scientists, and development teams to get data science workloads deployed to production.
MLOps will provide tuning, testing, and deployment of machine learning models including version control and partial rollout and rollback. These new capabilities will be particularly attractive to a CTO, CIO, data teams, and Ops teams.
Data, ML, and AI will make software much more useful. They also will require a ton of tooling. A great challenge for MLOps is the exploding toolchain of data science frameworks, vendors, and platforms. This is a problem GitLab has experience solving as it happened with DevOps.
We want to partner and integrate with existing players in the space to help unify and centralize data science workloads around organizations' existing DevOps and Source Code Management solutions. We believe GitLab provides a great platform to stitch together existing toolchains around existing production software development life cycles, bringing data science workloads closer to production.
Initially, our focus is on basic quality of life improvements to allow basic data science workloads on top of GitLab's existing CI/CD and runners. This will lay the framework for our AI Assisted group to build data science features on top of existing GitLab features enabling us to dogfood any DataOps and MLOps features we build.
We've started small by enabling CI jobs to leverage the GPU of GitLab runner with 13.9. This will allow us to explore integrating with widely used data science tools like Jupyter notebooks, open-source ML and AI frameworks, and data workflows.
Below are a few sample projects using GitLab CI and ML:
More reference implementation ideas can be found in the MLOps with GitLab overview.
Last Reviewed: 2022-10-13
Last Updated: 2022-10-13