With the self-service features of Domino Data Science Platform, you can keep your data science team productive by allowing them to select the language, libraries, and workspace type that best fit their requirements. Model deployment is handled inside the project, allowing one-click deployment of new versions. IT administrators retain the controls that they need with enterprise-grade identity management through Keycloak and role-based access control to delegate permissions.
Using the Design for Domino Data Science Platform, we have demonstrated an architecture capable of providing access to 304 Intel Xeon CPU cores across a default compute pool and three NVIDIA V100 GPGPUs in the GPU pool. For networking, we recommend 25 GbE using the Dell Networking S52xx series of switches. This white paper included information about two major Kubernetes distributions and their ability to support the features of Domino Data Science Platform.