RiseML

RiseML GmbH was a Berlin-based company founded in 2016 by Henning Peters and Elmar Haussmann. The firm developed a platform to streamline deep learning experiments on Kubernetes, targeting machine learning teams. Its core product was designed to abstract the computational resources of individual GPU servers, effectively treating them as a single, shareable cluster for training machine learning models. This allowed data scientists and engineers to concentrate on research rather than on managing infrastructure.

The platform provided a command-line interface (CLI) for users to prepare, execute, monitor, and scale their experiments in parallel across various machine learning frameworks like TensorFlow and Keras. It facilitated advanced techniques such as hyperparameter optimization and distributed training. A key aspect of the service was its flexibility; it could be installed on existing Kubernetes clusters, whether on-premise or in the cloud, offering users control over data privacy, security, and hardware choices. The business model was subscription-based, with a free Community Edition for individual users, a Professional Edition for teams starting at $249 per month, and a customizable Enterprise Edition.

In 2016, RiseML secured an undisclosed amount of seed funding from investors including Paua Ventures, German Accelerator, Fly Ventures, and TechCode Accelerator. The company announced its 1.0 release in January 2018. Despite its efforts to simplify the machine learning workflow, the company eventually ceased operations and is now considered deadpooled as of March 2019.

Keywords: deep learning platform, Kubernetes, machine learning operations, MLOps, GPU cluster management, model training, AI development, command-line interface, hyperparameter optimization, distributed training, on-premise AI, cloud AI, TensorFlow, Keras, data science infrastructure, Paua Ventures, German Accelerator, Fly Ventures, Henning Peters, Elmar Haussmann, Berlin startup