IBM Research |  CodeFlare
Digital tunnel made of tiny points of light
CodeFlare

Machine Learning. Distributed and Serverless. Build, scale, and unify entire pipelines. Effortlessly.

Try CodeFlare

Let your AI/ML teams focus on innovation. CodeFlare handles the infrastructure.

  • CodeFlare is an open-source, cloud-native platform for next generation AI and ML development

  • Minimize the effort and skills needed to scale your AI and ML

  • CodeFlare is built on OpenShift: deploy anywhere, from on-prem to cloud, and integrate easily with other cloud-native ecosystems.

CodeFlare dashboard screenshot
Use case
Highlight: Foundation models with CodeFlare

CodeFlare accelerates the automation of transfer learning tasks for foundation models. Today, businesses can use it to fully automate their foundation model pipelines. And with the benefits of hybrid cloud, your data never has to leave a secure environment.

Learn more →

Learn more

An Introduction to CodeFlareUpdate: CodeFlare is now refined for foundation modelsWhat are foundation models?CodeFlare Example: Scaling Explainable AICodeFlare Example: Tuning hyper-parametersCodeFlare Example: Fitting and scoring multiple classifiersFrom Ray Summit '21: Scaling and Unifying SciKit Learn and Spark Pipelines with RayFrom Ray Summit '21: Anomaly Detection on Remote Sensing with Ray + HorovodFrom Ray Summit '21: Powering Open Data Hub with Ray
Try CodeFlare
Curious? Request a demo, or sign up to be considered for our beta trial

If selected, you’ll get access to:

  • CodeFlare, with free GPU time
  • the research team that worked on CodeFlare
  • coding examples to help you get started
Apply →