By clicking on "Accept", you're agreeing to our privacy and cookie policy.

Data Version Control - Machine Learning Time Travel

Instead of reinventing the wheel, pairs AI with software development to create technology with data scientists, ML engineers, and data engineers in mind. We build new technology that helps alleviate challenges faced by AI teams to improve workflow and experience.

The process can seem simple - a data sample is made with a goal and features are defined. Next, you clean it up and your setup is ready to go. Once you process it through the machine-learning algorithm, you’ll receive results to display. Successful results might mean it’s break time, right? Not so fast…

Maybe a couple of days later, a colleague may ask, “Have you cleaned this feature from the noise?” EEK, it looks like you’ve overrode something. With a couple dependencies missing, the old file you previously used may not be so great. Well, what now? Fortunately, since you have a log where you’ve written down all of your steps, you and your colleagues can pick the experiment up at any point in the process. However, processing is going to cost you precious time that many don’t have or want to lose. Don’t worry! We have a fix for you.

Data Version Control (DVC)

Data Version Control (DVC) is an open-source data versioning tool that helps you define your pipeline regardless of what language you use. Data Version Control is based on Git, supports files of any size. Like a time machine, Git reproducibility can allow you to “warp” to any prior stage of your project without model retraining, and the best part - anyone can use it!

With Data Version Control, the process of fixing mistakes suddenly becomes incredibly easy. You’ll be able to share versions, fix problems, and collaborate in a much simpler way than ever before.

If you are looking for a straightforward method to define your pipeline and create a better way to share and ease into data science driven projects, consider Data Version Control.

DVC is among the many developer tools available through Iterative AI to create a streamlined process to assist AI teams and engineers through technology challenges. We are proud to have an impressive team behind us that are committed to providing secure and efficient solutions for increased productivity in data science.