Interactive
This is the same architecture from the front page, wired up and actually running. Turn up the traffic, starve the cache, slow the model down, kill something mid-flight. Poke it however you like and watch how the design responds. If you find a way to genuinely take it down, tell me which one .
A queue turns a latency problem into a capacity problem. More workers is more capacity.
Off, and every read pays a trip to the store instead.
A slow dependency is worse than a dead one. The breaker turns slow into fast-and-degraded, on purpose.
Or skip ahead
six things worth seeing · each card sets the controls above and tells you what to watch
Pick a scenario above. The caption under the diagram narrates what the system is doing; the numbers underneath say what it is costing.
This is a model, not a benchmark. The rates, latencies and limits are the model's own and are not measurements of anything in production. It is here because the behaviour is real even when the numbers are invented: queues grow, ceilings bind, and slow dependencies take systems down.
Off this page
the lab is a model · these are not
Personal project
An AI-powered intelligence layer for teams who need to stay on top of a set of companies and currently do it by hand. Specifics stay private until it's ready to be named.
Happy to walk through it in a demo.
Something for daily use. The build is essentially done; it goes to private testing once it holds up under real use.