TalksEngineering in production30 minEarlier version
Engineering Without a Safety Net: Where It Works and Where It Hurts
Recording on YouTube
An earlier version of this talk.
The talk has moved on: book Engineering Without a Safety Net: Where It Works and Where It Hurts.
The premise
Testing, monitoring, observability. They're rarely urgent, often skipped, and easy to dismiss when things are moving fast. But what's the real cost of deferring them, and how much can you actually get away with? This talk explores the tradeoffs of skipping traditional engineering practices in the name of speed, and what happens when you try to layer them in after a system has already scaled. It's not about dogma or checklists, it's about pragmatism, risk, and timing.
Who it's for
Engineers in fast moving teams shipping under pressure
You leave with
- Speed is a choice, risk is the bill you pay later
- When skipping safety nets is rational and when it is reckless
- Why retrofitting observability is harder than adding features
- False confidence is more dangerous than no confidence
- Engineering practices must match product maturity
- Reliability is an organisational problem, not just a technical one
How this talk evolved
- PHP Tek 2027 CFP
Engineering Without a Safety Net: Modernizing a 13-Year-Old System Without a Rewrite
PHP Tek 2027 submission cut of Engineering Without a Safety Net. Reframes the case study for mature PHP/Laravel and mixed-stack maintainers, emphasizes mapping inherited risk, prioritizing customer journeys and observability, adding safeguards without a rewrite, and expands the intended session to 50 minutes. Original and JSWorld 2027 versions preserved.
- JSWorld 2027 CFP
Engineering Without a Safety Net: Where It Works and Where It HurtsCurrent
Evolves the original discussion of deferred engineering practices into a production case study of a 13-year-old bootstrapped SaaS serving tens of millions of monthly users and generating tens of millions in annual revenue. Traces how reasonable decisions under earlier constraints ripple and compound into today's architecture and operational risk. Strengthens the focus on observability, critical user journeys, incident diagnosis, recovery verification, and prioritising incremental safeguards while continuing to ship. Title and abstract match the JSWorld 2027 CFP submission.
- 2025
Engineering Without a Safety Net: Where It Works and Where It Hurts