Facing the AI era, based on cloud computing technology, an unified management platform for enterprise’s AI computing resources, computing framework and algorithm application.

When you deploy a private AI service, you may face …

• Purchase a variety of models of servers, equipped with multiple models of CPU/GPU
• Depending on the business needs, multiple compute frameworks, such as TensorFlow, Caffe, and MXNet, need to be installed on different servers, and multiple versions need to be maintained at the same time
• Training sample data needs frequent copy management, training models and applications of multiple algorithms isolated from each other
• Assign and adjust resources for each training task, such as time slices, servers, and more
• Troubled in the details of the use of ultra-parametric tuning

We offer private/hybrid cloud solutions:

• Unified Management Enterprise’s Server/CPU/GPU/memory/Storage Resources, cross-geographic, cross-departmental, cross-model 
• Unified second-level deployment of multiple, multi-version computing frameworks to allocate computing resources on demand
• Unified portal, unified Roles Management
• Server bare Metal Start one-click Deployment, visual monitoring, scheduling cluster
• Very low efficiency loss based on container cloud architecture
• Cluster size Ultra fast elastic scaling

The platform can manage the unified management and resource allocation of different models of GPU/server/storage equipment, support the unification and simultaneous deployment of multiple, multi-version computing framework (TensorFlow/MXNet/Caffe, etc.), and support the application of unified management, storage service docking, ultra-parametric tuning and advanced functions of the algorithm.

• Based on K8S/Micro service/Docker, high performance, high elasticity, high availability of public and private hybrid deployments, the cost loss is extremely low.
• Seamless docking with device access platform and big data platform, can be deployed in PaaS/SaaS form.

Feature Introduction: