Design Patterns Roadmap

🧩 Design Patterns Roadmap: code → service → system → AI

How patterns actually make you better

Patterns are a vocabulary for trade-offs, not templates to paste. For each one, learn: the problem (forces), the structure, where it already exists in Spring/Go stdlib/Kafka/K8s, when NOT to use it, and implement it once in Java and once in Go.

The pattern ladder

Level	Note	Weeks
1. Object design (GoF + SOLID)	GoF Patterns (Java & Go)	W1–W4
2. Concurrency	Concurrency Patterns	W2–W3, revisit W10
3. Application & domain	Enterprise & DDD Patterns	W5–W9
4. Distributed & cloud	Distributed & Cloud Patterns	W10–W18
5. AI & agents	AI & Agent Patterns	W19–W22
Always	Anti-Patterns & Code Smells	every code review

Weekly pattern kata (every Saturday, 90 min)

1. Pick 2 patterns from this week’s list
2. Implement each in Java and Go (≤ 60 lines each), with a test
3. Find it in the wild: locate the pattern in Spring/JDK/Go stdlib/Kubernetes source and link the file in your note
4. Apply it to Orbit (or explain why it doesn’t fit)
5. Write “When NOT to use” in 3 bullets
6. Log it in the table below

Week	Patterns	Found in the wild	Orbit usage	Done
W1	Strategy, Factory	Comparator, BeanFactory	LLM provider adapters	[ ]
W2	Builder (+ Go functional options), Observer	RestClient.builder(), ApplicationListener	Workflow builder, run events	[ ]
W3	Decorator (Go middleware), Adapter	http.Handler chains, HandlerAdapter	Gateway middleware	[ ]
W4	State, Command, Template Method	Thread.State, Runnable, JdbcTemplate	Step state machine, step commands	[ ]
W5	Proxy, Chain of Responsibility	Spring AOP, FilterChainProxy	Tool auth chain	[ ]
W6	Repository, Unit of Work, Specification	Spring Data, Hibernate Session	Workflow repo	[ ]
W7	Composite, Visitor → sealed + pattern matching	ASTs, FileVisitor	Workflow DAG, expression AST	[ ]
W8	Interpreter, Memento	SpEL	orbit-expr, run checkpoints	[ ]
W9	Ports & Adapters, Anti-corruption layer	—	Provider/tool integrations	[ ]
W10	Lease, Fencing token, Idempotent receiver	etcd leases, Kafka producer IDs	Task leasing	[ ]
W11	Outbox, Inbox, Competing consumers	Debezium	Event backbone	[ ]
W12	Saga, Compensating transaction, Process manager	Temporal	UC3 refunds	[ ]
W13	Leader election, Scheduler-agent-supervisor	K8s leader election	Engine timers	[ ]
W14	CQRS, Materialized view, Event sourcing	—	Run history, search	[ ]
W15	Sidecar, Ambassador, Operator	Envoy, controllers	Worker pool operator	[ ]
W16	Gateway offloading, Valet key, Claim check	Envoy, S3 presigned URLs	Uploads, large payloads	[ ]
W17	Queue-based load leveling, Throttling, Bulkhead	KEDA, resilience4j	Workers, per-tenant caps	[ ]
W18	Pipes & filters, Deployment stamps/cells	Unix pipes	Ingestion pipeline	[ ]
W19	Prompt chaining, Routing, Model cascade	—	UC1, UC8, gateway	[ ]
W20	RAG variants, Reranking, Semantic cache	—	UC2	[ ]
W21	ReAct, Plan-and-execute, Orchestrator–workers, Supervisor	LangGraph	UC4, UC11	[ ]
W22	Evaluator–optimizer, Guardrail sandwich, Dual LLM, HITL	—	UC3, UC7, UC12	[ ]

Books

• Head First Design Patterns 2e ⭐ · refactoring.guru ⭐ · Design Patterns (GoF, reference)
• Patterns of Enterprise Application Architecture (Fowler) · Learning Domain-Driven Design (Khononov)
• ⭐ Patterns of Distributed Systems (Unmesh Joshi): WAL, leases, high-water mark, quorum, generation clock
• Microservices Patterns (Richardson) · Enterprise Integration Patterns (Hohpe) · Cloud Design Patterns (Microsoft Azure Architecture Center, free)
• Concurrency in Go · Java Concurrency in Practice · Anthropic Building effective agents