AI Application Architecture: Patterns for Building Intelligent Systems (Merced Books on Agentic AI and Data)

The architecture patterns that make AI systems production-ready.Building an LLM prototype takes hours. Designing an AI system that scales, stays reliable, and serves thousands of users requires architecture. This book provides the patterns.AI Application Architecture is a focused guide to the design patterns and architectural decisions that determine whether AI systems succeed or fail at scale. Written by Alex Merced, Head of Developer Relations at Dremio, it covers the structural patterns that bridge the gap between AI demos and production systems.What you will learn:Design request lifecycle architectures for synchronous and asynchronous AI inferenceImplement gateway patterns with model routing, fallback chains, and circuit breakersBuild multi-model architectures that route requests to the right model for each taskDesign RAG architectures with retrieval pipelines, context assembly, and generation layersImplement agent orchestration architectures with tool management and state machinesBuild event-driven AI systems for asynchronous processing at scaleDesign caching architectures that reduce costs and improve latencyImplement observability patterns for distributed AI systemsApply data access patterns using semantic layers and query federation through DremioChoose between monolithic, microservice, and serverless deployment topologiesImplement security patterns including authentication, authorization, and PII handlingWritten with current model specifications and real-world deployment patterns as of March 2026.For software architects: Design patterns specifically adapted for AI system requirements.For AI engineers: The structural foundation for scaling beyond prototypes.For platform teams: Reference architectures for shared AI infrastructure. Read more