https://www.davidlang.tech/tags/architecture/ Recent content in Architecture on David Lang Hugo en Mon, 01 Dec 2025 00:00:00 +0000 https://www.davidlang.tech/posts/from-rag-to-agentic-ai-whats-next-for-llm-powered-apps/ Mon, 01 Dec 2025 00:00:00 +0000 https://www.davidlang.tech/posts/from-rag-to-agentic-ai-whats-next-for-llm-powered-apps/ <p>The industry moved from chatbots → RAG → agents. Understanding the progression helps you invest in the right layer for your product maturity.</p> <h2 id="rag-era">RAG Era</h2> <p>Ground models in private data. Mature patterns: chunking, hybrid search, citations. Still the right default for Q&amp;A and search.</p> <h2 id="agent-era">Agent Era</h2> <p>Models call tools, plan multi-step workflows, and maintain state. Higher capability, higher risk.</p> <h2 id="whats-next">What&rsquo;s Next</h2> <ul> <li><strong>Evals-as-code</strong> in every pipeline</li> <li><strong>Smaller specialist models</strong> routed by orchestrators</li> <li><strong>On-device</strong> for privacy-sensitive steps</li> <li><strong>Human-agent collaboration</strong> UIs, not just chat</li> </ul> <h2 id="migration-path">Migration Path</h2> <p>Master RAG and evals first. Add one well-scoped agent tool. Measure task completion before expanding autonomy.</p> https://www.davidlang.tech/posts/building-multi-agent-ai-systems/ Tue, 20 May 2025 00:00:00 +0000 https://www.davidlang.tech/posts/building-multi-agent-ai-systems/ <p>Multi-agent systems divide work among specialized agents-a researcher, coder, critic-coordinated by a supervisor or message bus.</p> <h2 id="patterns">Patterns</h2> <p><strong>Supervisor</strong> - One model delegates subtasks and aggregates results.</p> <p><strong>Peer-to-peer</strong> - Agents message each other until consensus or max rounds.</p> <p><strong>Pipeline</strong> - Fixed stages (plan → implement → test).</p> <h2 id="implementation-tips">Implementation Tips</h2> <p>Give each agent a narrow system prompt and tool set. Pass structured state (JSON) between agents, not raw chat logs.</p> <h2 id="failure-modes">Failure Modes</h2> <p>Infinite loops, duplicated work, conflicting edits. Enforce step limits, idempotent tools, and single-writer rules for shared files.</p> https://www.davidlang.tech/posts/building-microservices-with-nodejs/ Sat, 22 May 2021 00:00:00 +0000 https://www.davidlang.tech/posts/building-microservices-with-nodejs/ <p>Microservices split a system into independently deployable services. Node.js is a common choice for I/O-bound services, BFF layers, and event-driven workers.</p> <h2 id="service-boundaries">Service Boundaries</h2> <p>Split by business capability-not by technical layer. A <code>orders</code> service owns order data; a <code>notifications</code> service sends email. Communicate via REST, gRPC, or message queues (RabbitMQ, Kafka).</p> <h2 id="resilience-patterns">Resilience Patterns</h2> <p>Implement timeouts, retries with exponential backoff, and circuit breakers. Use correlation IDs across services for distributed tracing (OpenTelemetry).</p>