Writing
Field notes from backend architecture audits and scaling work.
An internal SaaS Reliability Lab shows why small products usually fail on sync, auth drift, background jobs, and silent state corruption before they fail on raw traffic.
A real incident on eliasfeijo.dev: privacy extensions blocked analytics and the scheduler handoff, turning a booking CTA into a silent funnel failure until I fixed both the architecture and the UX.
Retry composition, circuit breaker traps, and the observability mistakes that turn small failures into expensive ones — validated in 2 days with a spec-first, AI-assisted workflow.
A 12-commit build log of a Go reliability laboratory — retry composition, circuit breaker gotchas, and the patterns that make failures bounded and recoverable.
Why small SaaS systems break after launch — and the hidden reliability problems founders usually discover after their first real users.