Engineering Without a Safety Net: Where It Works and Where It Hurts

What happens when a SaaS product bootstrapped by a couple of people grows into a 13-year-old business serving tens of millions of users per month and generating tens of millions in annual revenue?

You inherit more than a codebase. You inherit years of decisions made under different constraints. Shortcuts that kept the business alive become dependencies. Assumptions outlive the people who made them. Individually reasonable choices interact and compound into a system that nobody would design from scratch, but that customers depend on every day.

You cannot pause the business to retrofit every engineering practice a well-funded team might have established from the beginning. So how do you decide where to start?

This talk follows the consequences of those decisions through architecture, testing, observability, and production operations. We’ll explore how to assess risk when visibility is incomplete, distinguish tolerable technical debt from dangerous blind spots, and choose where safeguards will make the greatest difference.

Observability is central to that work. What do healthy dashboards actually tell us? Which user journeys need stronger signals? When an incident happens, can we understand its impact, trace the failure, and verify that recovery worked? We’ll look at how those questions guide investment in instrumentation, testing, and safer releases while features still need to ship.

You’ll leave with a practical way to navigate an established system: understand why it became what it is, make accumulated risk explicit, and improve it incrementally without assuming the team, budget, or clean slate you wish you had.