Skip to main content
Version: 1.0.0

Key Features

This page introduces the key features of the NuFi platform.

NPUOps​

From Existing MLOps to NPU Serving

Bring in models trained in existing MLOps environments such as Kubeflow and MLflow, then handle NPU porting (compilation → validation → model registry publication) and model serving.

Supported devices: GPU (NVIDIA), RNGD (FuriosaAI).

Lab​

When You Need a Training Environment

Create web-based development environments such as Jupyter Notebook, VS Code, and LlamaFactory. Models trained in a Lab can move directly into NuFi's serving flow.

See Lab for details.

Security​

RBAC and Audit Logs

Separate administrator and user actions with Keycloak-based role control. Audit logs record who performed each action, when it happened, and what changed, so they can be used for security reviews.

  • RBAC: Separate project, user role, and administrator permissions to restrict access to sensitive features.
  • Audit logs: Record user and administrator activity for change tracking and security audits.

Evaluation Dashboard​

GPU vs NPU

Deploy the same model to an NPU and a GPU, then compare accuracy, TPS, TTFT, and power efficiency (tokens per watt). Save evaluation data and criteria to repeat checks under the same conditions, and choose the artifact that best matches your operating goal from multiple NPU compile results.

See the dataset, evaluation, and compile option optimization guides for details.

Integrated Hardware Monitoring​

Cluster and Device Status

From the cluster down to individual devices, observe everything inside the platform UI.

  • Cluster-wide and per-node CPU · memory · storage status
  • Per-device utilization, temperature, power, and memory visualization
  • Devices in the "In Use" state directly show the occupying Pod name — instantly trace the source when an incident occurs

See Monitoring for details.