A curated collection of real-world production engineering patterns
Backend · DevOps · Data · Systems · ML · Security
Learn how Netflix, Google, Uber, and Stripe build production systems.
Contribute patterns, fix docs, or add new modules.
This repo is a living reference of production-grade engineering patterns. Each pattern includes:
- Real-world architecture — how top tech companies solve the problem
- Code examples — runnable snippets in Python, Go, YAML, SQL
- Trade-off analysis — when to use (and when not to use) each pattern
- Contribution-ready — every sub-pattern is independently extendable
Think "public-apis" but for engineering systems.
| Pattern | Description | Reference |
|---|---|---|
| API Gateway | Gateway, BFF, Service Mesh patterns | Netflix Zuul, Google ESP |
| Microservices | Service decomposition, gRPC, observability | Uber, Spotify |
| Event-Driven | Kafka/Pulsar, event sourcing, DLQ | LinkedIn Kafka, Stripe |
| CQRS | Command Query Separation, read models | AWS, Event Store |
| Circuit Breaker | Resilience patterns, bulkheading | Netflix Hystrix, Amazon |
| Idempotency | Exactly-once, idempotency keys | Stripe API |
| Pattern | Description | Reference |
|---|---|---|
| Kubernetes Patterns | Sidecar, operator, HPA, network policies | Google Borg, OpenAI |
| CI/CD Pipelines | GitOps, build caching, artifact promotion | Netflix Spinnaker |
| Terraform Modules | IaC, state management, policy-as-code | Airbnb, HashiCorp |
| Monitoring | USE/RED, Prometheus, tracing | Google SRE, Netflix Atlas |
| Canary Deployments | Traffic splitting, auto-rollback | Netflix Kayenta |
| Chaos Engineering | Gamedays, blast radius, Litmus | Netflix Chaos Monkey |
| Pattern | Description | Reference |
|---|---|---|
| Streaming Pipelines | Flink/Kafka Streams, watermarks, state | Netflix Keystone, Uber AthenaX |
| Batch Processing | Spark/Dask, partitioning, broadcast joins | Google MapReduce, Databricks |
| Data Lake | Iceberg/Delta/Hudi, medallion architecture | Netflix Iceberg, Uber Hudi |
| Data Warehouse | Star schema, MPP, columnar storage | Snowflake, BigQuery |
| Schema Evolution | Avro/Protobuf, compatibility, registries | LinkedIn Schema Registry |
| Pattern | Description | Reference |
|---|---|---|
| Distributed Cache | LRU/LFU, write strategies, Redis cluster | Netflix EVCache, Twitter |
| Leader Election | etcd/ZK, lease-based, K8s pattern | Google Chubby, etcd |
| Distributed Queue | Kafka partitioning, SQS, backpressure | LinkedIn Kafka, Amazon SQS |
| Consistent Hashing | Virtual nodes, ring hashing, rebalancing | Amazon Dynamo, Discord |
| Consensus | Paxos, Raft, CAP, split-brain | Google Chubby, etcd |
| Pattern | Description | Reference |
|---|---|---|
| Feature Store | Online/offline, point-in-time, Feast | Uber Michelangelo, Tecton |
| Model Serving | Triton/BentoML, canary, autoscaling | Netflix Meson, Google Vertex |
| ML Pipelines | Kubeflow/Metaflow, HPO, artifact tracking | Netflix Metaflow, Google TFX |
| Drift Detection | PSI/KS, adaptive thresholds, retraining | Uber, Netflix, Google |
| A/B Testing | Experiment design, Bayesian, sequential | Google Overlapping, Netflix |
| Pattern | Description | Reference |
|---|---|---|
| Authentication | OAuth 2.0, OIDC, JWT, MFA, WebAuthn | Google Auth, Auth0 |
| Authorization | RBAC/ABAC/ReBAC, OPA, Zanzibar | Google Zanzibar, AWS Cedar |
| Secret Management | Vault, dynamic secrets, sealed secrets | Netflix Conjuer, HashiCorp |
| Endpoint Security | Rate limiting, WAF, DDoS, API security | Cloudflare, Stripe |
| Audit Logging | Hash chains, CloudEvents, compliance | Stripe, Google |
This is a reference repository — no build needed. Browse patterns by module:
# Clone the repo
git clone https://github.com/YOUR_ORG/awesome-production-engineering.git
# Pick a module and start learning
cd awesome-production-engineering/backend-systems/circuit-breaker
cat index.md # Read the pattern
cd examples
python circuit_breaker.py # Run the codeWe follow the public-apis contribution model:
- Pick a pattern from an existing module or propose a new module
- Each sub-pattern must have:
index.md+examples/with runnable code - Include real-world references and trade-off analysis
- Open a PR — maintainers review within 48h
Full Contribution Guide → | Good First Issues → | Code of Conduct →
-
index.mdexists with diagrams and references -
examples/has runnable code - Real-world company reference included
- Trade-offs section: "when to use" vs "when not to use"
- Mermaid diagram (where applicable)
| Metric | Value |
|---|---|
| Modules | 6 |
| Patterns | 31 |
| Code Examples | 50+ |
| Lines of Code | 15,000+ |
| Languages | Python, Go, YAML, SQL, Rego, Terraform |
| Real-world References | 30+ companies |
- For interviews: Browse
system-design/— covers all major distributed systems patterns - For architects: Start with
backend-systems/andsecurity-patterns/ - For ML engineers: The
ml-systems/module covers the full ML lifecycle - For DevOps:
devops-infrastructure/is your go-to - For data engineers:
data-engineering/has production data stack patterns
MIT — see LICENSE