Architecture overview¶
Layers¶
| Layer | Contents |
|---|---|
| Platform services | Internal OCI registry; Istio and Gateway API |
| Serving | Per-model vLLM stacks orchestrated by llm-d, behind an inference gateway |
| Application | shaide server (API, auth, users) and the control panel |
| Packages | Installer, observability stack, shared libraries |
Each layer is an independent Pulumi program.
Deployment order¶
1. Internal OCI registry infra/cloud-harbor
2. Gateway + Istio infra/gateway-provider
3. Model serving app_serving
4. Application layer app_shaide
5. Optional add-ons app_mcp, monitoring
The installer performs all five.
Request path¶
client → ingress → shaide server → inference gateway → GAIE → model replica
Embedding requests go directly from the shaide server to the embedding service, bypassing the gateway.
Detail: Architecture.
