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.