feat: OpenSearch RAG demo

demos/demos-v2.yaml

@@ -286,3 +286,28 @@ demos:
       cpu: 6400m
       memory: 12622Mi
       pvc: 20Gi
+  opensearch-rag:
+    description: Retrieval Augmented Generation (RAG) demo with OpenSearch vector search and Llama 3.1 8B for semantic question answering
+    documentation: https://docs.stackable.tech/home/stable/demos/opensearch-rag
+    stackableStack: opensearch-rag
+    labels:
+      - opensearch
+      - rag
+      - vector-search
+      - ai
+      - llm
+      - llama
+      - ollama
+      - jupyterlab
+      - semantic-search
+    manifests:
+      - plainYaml: ./demos/opensearch-rag/load-embeddings-from-git.yaml
+    supportedNamespaces: []
+    resourceRequests:
+      cpu: 8000m
+      memory: 16Gi
+      pvc: 10Gi
+    parameters:
+      - name: branch
+        description: Git branch to download demo files from
+        default: main

demos/opensearch-rag/embeddings/README.md

@@ -0,0 +1,55 @@
+# OpenSearch RAG Demo - Embeddings Generation
+
+This directory contains the script to regenerate embeddings for the Stackable documentation.
+
+## Prerequisites
+
+**Ollama with nomic-embed-text:v1.5 model** - Choose one of the following options:
+
+### Option 1: Run Ollama locally
+
+```bash
+# Install Ollama
+curl -fsSL https://ollama.com/install.sh | sh
+
+# Pull the embedding model
+ollama pull nomic-embed-text:v1.5
+```
+
+### Option 2: Port-forward from the demo
+
+```bash
+# Deploy the demo if not already running
+stackablectl demo install opensearch-rag
+
+# Port-forward Ollama (keep running in a separate terminal)
+kubectl port-forward deployment/ollama 11434:11434
+```
+
+## Regenerate Embeddings
+
+Run the script using uv to automatically handle Python dependencies:
+
+```bash
+uv run --with requests --with beautifulsoup4 python generate-embeddings-local.py
+```
+
+This will:
+- Crawl docs.stackable.tech (defaults to `stable` release)
+- Extract and chunk documentation
+- Generate embeddings using nomic-embed-text
+- Save to `stackable-docs-embeddings.json` (~89 MB, ~4200 chunks)
+
+The process can take more than 1 hour mainly depending on your CPU.
+
+### Configuration
+
+You can customize the crawling behavior with environment variables:
+
+```bash
+# Crawl a specific Stackable release (default: stable)
+RELEASE=24.11 uv run --with requests --with beautifulsoup4 python generate-embeddings-local.py
+
+# Crawl nightly documentation
+RELEASE=nightly uv run --with requests --with beautifulsoup4 python generate-embeddings-local.py
+```