Skip to main content

Workbench

AI/ML workspace — sessions, models, endpoints, training, pipelines, sweeps.

The Vantage AI Workbench

Workbench is where you build, train, and serve machine-learning models on Vantage. It bundles eight tools — sessions, models, endpoints, training jobs, pipelines, sweeps, compute profiles, and observability — into one workspace, all sharing the same compute, storage, and cost envelope.

info

Workbench replaces the Kubeflow dashboard. Everything you'd reach through Kubeflow Notebooks, KServe, Trainer, Katib, and the Pipelines UI is here under native Vantage primitives — without the namespaces, CRDs, and Istio routes leaking through.

What you'll find inside​

  • Sessions — Interactive notebook environments — Jupyter, VS Code, RStudio — pinned to GPU pools and your team's storage.
  • Models — Versioned model catalog. Pull from HuggingFace or your own training runs, then deploy to an endpoint.
  • Endpoints — Inference services with autoscaling, canary rollouts, and authenticated URLs. Predictive or LLM.
  • Training Jobs — Distributed training on PyTorch, DeepSpeed, or MLX runtimes. Retry, suspend, resume.
  • Pipelines — Multi-step DAGs that orchestrate ingestion, training, evaluation, and deployment.
  • Sweeps — Hyperparameter search with Bayesian, grid, or random algorithms — tracked end-to-end.
  • Compute Profiles — The reusable shape of your compute: GPU type, count, autoscaling bounds, instance class.
  • Observability — Cluster-wide rollups of utilization, spend, idle GPU hours, and live alerts.
⌘I