Library
Architecture blueprints
Reference architectures with their assumptions made explicit: what the shape is, what it costs, how it fails, and what has to be true before it is a reasonable choice. Read the assumptions first — they are what decide whether the rest applies to you.
AI Application
reviewed 2026-09-21A production feature built on a third-party model: version pinning, evaluation, cost and latency budgets, guardrails, and a defined answer for when the provider degrades.
Data Platform
reviewed 2026-09-21Ingestion, a warehouse and a semantic layer, run as a product: contracts at the boundaries, freshness that is measured, and quality tests that can actually stop bad data.
Event-Driven Architecture
reviewed 2026-09-21Asynchronous services coordinated through a durable log or queue — what it buys you, what it costs in observability and debugging, and the guarantees you must decide on up front.
High-Traffic API
reviewed 2026-09-21A read-heavy public API built for predictable behaviour under load: caching layers, backpressure, and the failure modes that only appear when the traffic is real.
Multi-Region Application
reviewed 2026-09-21Running in more than one region: what problem it actually solves, the data consistency choice you cannot avoid, and why most teams should exhaust single-region options first.
Production Kubernetes Platform
reviewed 2026-09-21A managed-control-plane Kubernetes platform sized for several teams: what the shape is, what it assumes about your operational maturity, and what it costs to run properly.
Retrieval-Augmented Generation System
reviewed 2026-09-21A system that answers from your own content: ingestion and freshness, embedding versioning, permission-aware retrieval, and the measurement that tells you which half is broken.
SaaS Startup Platform
reviewed 2026-09-21The deliberately boring architecture a small team can actually operate: one application, one managed database, and the smallest set of moving parts that still supports paying customers.