AI Agent Memory Persistence and State Synchronization Solution

You are an expert AI agent memory architect. I need you to design a comprehensive memory persistence and state synchronization system for my AI agent. ## Context - Agent type: [describe your agent - coding assistant / research agent / personal assistant] - Runtime environment: [local / cloud / hybrid] - Expected conversation volume: [sessions per day] - Memory backends available: [SQLite / PostgreSQL / Redis / Vector DB] ## Requirements Please design and provide: 1. **Memory Architecture** - Short-term memory (current session context) - Working memory (active task state) - Long-term memory (persistent knowledge) - Episodic memory (past interaction summaries) 2. **State Synchronization Strategy** - How to serialize/deserialize agent state between sessions - Conflict resolution when multiple sessions access shared memory - Incremental sync vs full snapshot tradeoffs 3. **Retrieval Strategy** - When to use semantic search vs keyword lookup - Memory relevance scoring and decay functions - Context window budget allocation 4. **Implementation Skeleton** - Provide code structure (Python or TypeScript) for the memory manager - Include interfaces for pluggable storage backends - Add memory compaction/summarization pipeline Output format: Structured technical design document with code examples.