本地文档语义搜索系统一键部署脚本生成器

You are a senior search infrastructure engineer specializing in semantic search systems. Based on my requirements, generate a complete local document semantic search deployment plan.

My Requirements

• Document types: [PDF/Markdown/HTML/Code/etc.]
• Total document count: [number]
• Average document size: [size]
• Hardware: [CPU/GPU/RAM specs]
• Privacy requirement: [fully local / hybrid]

Generate the following:

1. Architecture Decision

• Embedding model recommendation (with size/speed/quality tradeoffs)
• Vector database selection (Chroma vs Qdrant vs Milvus Lite vs LanceDB)
• Chunking strategy (size, overlap, semantic boundaries)

2. Docker Compose File

• Complete docker-compose.yml for the chosen stack
• Environment variables with sensible defaults

3. Ingestion Script

• Python script to process documents, generate embeddings, and store in vector DB
• Support for incremental updates

4. Search API

• FastAPI endpoint for semantic search with hybrid retrieval
• Reranking step for top results

5. Benchmark Commands

• Commands to test search quality and latency
• Expected performance baselines for my hardware

Be specific and production-ready. Include error handling and logging.