Technology - MX Momentum

Excellence

Our state-of-the art technology stack enables us to efficiently build and operate production grade ML systems.

in AI

technology

Enhancing AI research,

development

and operations

Data

In AI it’s all about data! Building high quality datasets is important to build a good model.

GO TO DATA

Model

Understanding a problem, designing and building the right model is an art of its own.

GO TO MODEL

Deployment

Now, bring it to production! Many AI-prototypes never reach production since deploying AI systems has its own challenges.

GO TO DEPLOYMENT

01 Data

Data is the fuel of AI. High quality data is important for the success of an AI project. We developed processes and tools for data-centric AI and creating valuable insights. Our project teams face various data challenges when training models. Having the right tools at hand significantly improves the outcome.

That’s why we created Parrot to leverage domain experts for solving NLP tasks. Moreover, every day we are pushing the state-of-the art of ML which requires complex data infrastructure. Staying flexible at scale is a task for Squirrel!

Squirrel

Squirrel is a data infrastructure library which enables ML teams to share, load, and transform data in a collaborative, flexible, and efficient way. And it's open source!

CHECK IT OUT

Parrot

Parrot is a framework to build custom-made process workflows, comprising labeling of text documents. You’ll never encounter our parrot in the wild, but you may meet its descendants. Which ones?

CHECK IT OUT

02 Model

Fueled by data, ML models are the mathematical engines of AI. To build them you need to understand the problem and come up with an appropriate algorithm to model your task.

We combine this artisan art with automatization of frequent steps. Each project needs a Scaffold that holds it together. Our Chameleon framework helps us to bring state-of-the art computer vision models into production in no time!

Chameleon

Chameleon is bootstrapping all our computer vision (CV) projects with battle-tested models and code. It is our framework for implementing generic tasks like data preprocessing or data augmentations for CV. It includes experiment tracking, hyperparameter optimization and a large computer vision model zoo.

CHECK IT OUT

Scaffold

Scaffold is our project agnostic and opinionated tool box which provides utilities and interfaces in order to accelerate our development. It builds on existing technologies, enables us to share our learnings as reusable code and covers the whole ML lifecycle.

CHECK IT OUT

03 Deployment

The ML lifecycle does not end but begins with the model deployment step: This is when the system starts generating value! At Momentum, we deploy your ML models at scale and operate them in a fully managed fashion, providing continuous third level support. We can monitor the models’ behavior in production in order to ensure that they reliably solve your use case.

ML Ops

Many Data Science teams struggle with bringing PoCs to production. MLOps best practices greatly increase the number of projects that make this transition while drastically reducing the systems’ maintenance cost.

CHECK IT OUT