Implement Fastify vector embeddings and semantic search demo

[Copilot]

Builds a working demonstration of vector embeddings and semantic similarity search using Node.js and Fastify.

Implementation

Core modules:

• embeddings.js - Text-to-vector conversion using character/word features (10-dimensional), cosine similarity, and Euclidean distance
• vectorStore.js - In-memory vector store with similarity search and CRUD operations
• server.js - Fastify REST API with 9 endpoints for document management and search

API endpoints:

POST /documents    # Add document with auto-generated embedding
POST /search       # Semantic search with cosine or euclidean similarity
GET  /documents    # List all documents
GET  /stats        # Store statistics
POST /demo         # Load sample data

Example usage:

# Load demo data
curl -X POST http://localhost:3000/demo

# Search for similar documents
curl -X POST http://localhost:3000/search \
  -H "Content-Type: application/json" \
  -d '{"query": "programming languages", "topK": 3}'

Testing:

• test.js - 6 test cases covering embedding generation, similarity calculations, and store operations
• Run via npm test

Notes

• Embeddings use simplified text features for demonstration; production would use pre-trained models (OpenAI, Cohere, sentence-transformers)
• CORS configured with origin: true for demo; restrict in production
• Requires Node.js ≥18.11.0 for --watch flag support

Original prompt

node.js fastify template application, logic for demo of vector embeddings and retrieval

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.