Browse All Patterns

14 patterns across 10 pillars. Each pillar covers a distinct layer of a production AI system. Click a pillar to see all its patterns, or explore the pattern graph to see how they connect.

Inference & Serving 3 patterns — Gateway routing, semantic caching, and model routers.

Data Patterns 1 pattern — Data contracts, feature stores, and training pipelines.

Reliability & Resilience 1 pattern — Circuit breakers, graceful degradation, and canary deployments.

Retrieval & Memory 1 pattern — Hybrid search, RAG, reranking, and contextual memory.

Observability 1 pattern — Span-level tracing, cost attribution, and quality drift detection.

Security & Trust 1 pattern — Input sanitization, output validation, and PII scrubbing.

Cost & Efficiency 1 pattern — Token budgets, prompt compression, and cost circuit breakers.

Governance 1 pattern — Model cards, data lineage, and policy-as-code.

Graph Patterns 3 patterns — GraphRAG, graph reasoning, and entity resolution.

Evaluation & Testing 1 pattern — LLM-as-Judge, eval pipelines, and regression testing.