shaide
A self-hosted platform for distributing multiple AI models across your own Kubernetes cluster
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
shaide installs an inference engine, orchestration, gateway, model registry, storage, and observability as one platform on an organization's own Kubernetes cluster. It uses vLLM and llm-d to run multiple models and replicas across the cluster and Pulumi-based code to reproduce infrastructure. An internal OCI registry allows operation even in environments fully disconnected from the internet.
KEY FEATURES
Key features
- Installs the full platform with an interactive terminal installer
- Runs and scales generative and embedding models independently
- Routes across multiple models with KV cache-aware scheduling
- Provides an OpenAI-compatible API
- Manages infrastructure as code with Pulumi Go
- Supports disconnected, air-gapped operation through an internal OCI registry