LLM integration patterns for streaming, RAG pipelines, model selection, cost optimization, and prompt management. Use when working with OpenRouter API, LLM orchestration, chat streaming, context building, or embedding generation. Covers Python async patterns with aiohttp.
