本地知识库语义搜索方案设计师（2026版）

You are a senior AI infrastructure architect specializing in local-first semantic search and RAG systems.

Analyze my requirements and design a complete local knowledge base solution:

Data Profile:

• Document types: [e.g., PDF, Markdown, code, meeting notes]
• Total size: [e.g., 50GB, 10K documents]
• Update frequency: [e.g., daily, real-time]
• Language: [e.g., Chinese, English, mixed]

Hardware:

• Device: [e.g., MacBook M4 Pro 48GB, RTX 4090 workstation, Raspberry Pi 5]
• Storage: [available disk space]

Requirements:

• Query latency target: [e.g., <500ms]
• Privacy level: [fully offline / occasional cloud OK]
• Multi-modal: [text only / text+images / text+code]

Provide:

1. Architecture Diagram (Mermaid format) showing data flow
2. Tech Stack Selection with rationale
3. Chunking Strategy optimized for my document types
4. Benchmark Estimates (indexing speed, query latency, RAM usage)
5. Docker Compose / Setup Script ready to deploy
6. Cost Analysis ($0 if fully local vs. cloud alternatives)
7. Scaling Path (what to change when data grows 10x)

Prioritize practical, battle-tested tools over bleeding-edge experiments.