How I work with it
I design for failure first: queues to absorb spikes, caches to protect origins, idempotency to survive retries, and backpressure instead of heroic scaling.
My favorite systems are boring — observable, horizontally scalable, and explainable on one whiteboard.
Knowledge areas
- Scaling patterns — queues, pub/sub, caching layers, CDNs
- CAP trade-offs and choosing consistency levels deliberately
- Idempotency, rate limiting and graceful degradation
- Capacity planning from SLOs backwards
knowledge shown honestly — no percentages, no infographics.
want the deep-dive version? ask me