ML Infrastructure
Our Continuous Integration System is designed to streamline and accelerate the process of fast and cost-effective AI development. By automating the integration of code changes, our system ensures efficiency and affordability throughout the development lifecycle.
Give your ML Development a Workflow
ML development will be faster than ever. Queue AI trainings, test you code before running it on the GPU, run tests and collect the results. And after all that, don't forget what you've done. All this requires only 4 steps.
Create a Project File
Start your Project with a flake.nix file. Read our Documentation on how to install Nix.
Configure Datasets and Training Pipeline
We configure all datasets and trainings declarative. This let us remember each training we did. You even can declare multiple models, datasets and tests at once. All configuration options available under https://search.wavelens.io
Push to Git
We only commit our Code and flake.nix file. We don't save entire models or datasets.
Let Hydra do the Rest
Hydra is a CI System, that compiles our derivations. Models that have missing dependencies will also be build. After we are done with the trainings process, we are able to download the results as a .zip file.
Want to know more?
Visit our Documentation page and find out how our ML Pipeline works.
Why is AI-Infrastructure needed?
We know that organizing ML Developments is hard. That's why we are here to help you to keep focusing on the development.
Many Devs - 1 Server
Servers have limited compute power. We distribute the load over time.
No Space?
Don't save dozens of models. Be reproducible and remember the models you've trained by only saving the code.
Setting up Servers?
Only spend money on servers that you use. No need to waste time setting up servers.
Different Versions?
Commit the state of your environment and let every team member work with the same versions and dependencies.
Interested?
Please don't hesitate to contact us, if you want to learn more about our Infrastructure.
Flawless server integration into our Infrastructure Process
Need more Compute Power? We provide AI Servers of all kinds.
Set up GPU resources for all of your AI training and deep learning needs, which are resource-intensive in terms of computing power.
FAQs
Common questions we get by our customers.
No. While we support those in our AI Infrastructure solution, we don’t solve any problems related to their solutions.
Yes. While we also offer a hosting solution, you can host our solution yourself.
Yes. We have a fair pricing plan and even do free plans.
Yes. You can connect your own machines to train your models. If you don’t have enough compute power, we are happy to offer our servers too.
We change your workflow to some degree.
For example we enforce creating a config.json with your models hyperparameters and we’ll get rid of Pypi or Conda, but after all we keep it as simple as possible to understand all changes we make to your workflow.
Yes. We support all possible packages in the Nix Store.
And if you are missing a package we help you by adding it to the Nix Store. Search for your required packages on https://search.nixos.org.
Get in touch with us!
Special Problems? Contact Us! We'll find a solution.
