apiVersion: catalyst.openova.io/v1alpha1
kind: Blueprint
metadata:
  name: bp-bge
  labels:
    catalyst.openova.io/category: ai-runtime
    catalyst.openova.io/section: pts-4-6-llm-serving
spec:
  version: 1.0.0
  card:
    title: BGE Embeddings + Reranker
    summary: BAAI General Embedding (sentence-transformers + bge-reranker). CPU-friendly multilingual embeddings + cross-encoder reranking. Default model bge-small-en-v1.5; bp-llm-gateway discovers via Service annotation.
    icon: bge.svg
    category: ai-runtime
    tags: [embeddings, reranker, rag, sentence-transformers, ai]
    documentation: https://huggingface.co/BAAI
    license: MIT
  visibility: listed
  owner:
    team: ai-platform
    contact: ai-platform@openova.io
  configSchema:
    type: object
    properties:
      embeddingModel:
        type: string
        default: "BAAI/bge-small-en-v1.5"
        description: HuggingFace model ID for the embeddings endpoint.
      rerankerModel:
        type: string
        default: "BAAI/bge-reranker-base"
        description: HuggingFace model ID for the reranker endpoint.
      enableReranker:
        type: boolean
        default: true
        description: Whether to start the reranker container alongside embeddings.
      replicas:
        type: integer
        default: 1
        minimum: 1
        maximum: 8
      maxBatchSize:
        type: integer
        default: 32
      maxLength:
        type: integer
        default: 512
        description: Token cap per request (bge-small-en-v1.5 supports 512; bge-m3 supports 8192).
  placementSchema:
    modes: [single-region, active-active]
    default: active-active
  manifests:
    chart: ./chart
  depends:
    - blueprint: bp-cnpg
      version: ^1.0
      alias: cnpg
  upgrades:
    from: ["0.x"]
  observability:
    metrics: prometheus
    logs: stdout