Forge Engineering
Opinionated engineering notes on building production SaaS platforms - from runtime architecture and platform engineering to the realities of running software at scale.
-
AMP Remote Write HTTP 400: The CloudWatch Metric That Solved It
Intermittent Amazon Managed Prometheus remote-write failures looked like encoding or signing bugs. The real signal was CloudWatch DiscardedSamples — and the fix was embarrassingly familiar if you've operated Prometheus at scale.
Read article → -
The software nobody plans to build - but every successful team eventually does...
Every software company starts with one product. Given enough time, almost every successful team ends up building a second one they never planned for.
Read article → -
What "production-ready" actually means - and why most teams discover it too late.
Production readiness isn't about whether a system runs. It's about how it behaves under failure, change, and scale - and most teams discover the gap too late.
Read article → -
The real startup killer isn't product - it's building platform foundations from scratch.
Most early-stage teams think they're building one product. They're actually building two - and the second one quietly consumes man-years of engineering time.
Read article →