Open Source Infrastructure as a Service

Simplified

What is Unikorn?

Unikorn is an opinionated suite of services that manage the provisioning and life-cycle of compute infrastructure. The goal is to abstract away all the tedious day-to-day tasks associated with running compute infrastructure.

We've built a system to be truly proud of, leveraging best-in-breed Open Source components, and combining this with enterprise standard security best practices. The end result is easy to use, and secure by default.

Flexible Security

Security is at the heart of everything in Unikorn.

We don't ever collect or manage user passwords or credentials. This functionality is delegated to Google, Microsoft (covering most enterprise users), and GitHub for others. You can even bring your own OIDC compliant identity provider if your organization isn't covered by the supported options.

On top of authentication, Unikorn support flexible role based access control (RBAC) where users are granted fine-grain access control to individual API endpoints. Most services act on behalf of a user reducing the permissions a user requires to actually provision infrastructure.

Learn More

Multi Region Cloud Providers

Unikorn is architected in a way where it's only concerned with servers, networks, security group rules, etc. These are abstractions of the underlying cloud provider implementation, so everything appears and behaves the same. A network, for example, will be ready to run any kind of workload, handling the HPC requirements of AI workloads for example.

Regions are defined as instances of a cloud provider e.g. an OpenStack deployment.

Learn More

Simple Cluster Provisioning and Life-cycle Management

Unikorn natively supports baremetal and Kubernetes cluster provisioning. The focus is on simplicity. You're only presented with limited options: cluster geometry and core software versions. This means less time procrastinating and more time doing what you actually need to do.

For Kubernetes clusters in particular, Unikorn provisions clusters with a known working set of applications (powered by Helm). These are bundled together into tested and stable releases, and automatically upgraded to provide security guarantees around CVE mitigation.

Learn More

Designed for AI/ML

Although Unikorn can be used for general purpose compute, its core focus will always be on AI. GPUs are first-class citizens, and the intention is always to provide higher level machine learning and generative AI services.

With that aim, Unikorn has been designed to support rapid 3rd party service integration with your AI/ML applications. All the authentication, authorization and monitoring capabilities are available with little effort through reusable libraries that integrate with Unikorn services.

Easy to Use

We've been inspired by the best designs in the world.

No one wants to read reams of documentation to understand how something works. We get you, and our mission is to provide simple and intuitive interfaces that anyone can use. We've reduced the input you need to provide to get up and running quickly to the bare minimum. Fewer distractions, more productivity.

We want Unikorn to be the tool we – as engineers – want to use. But where that's not sufficient, Unikorn provides a rich REST API to enable your creativity and provide automation.