Prague DevOps Meetup

Talk 1
Title: The Platform Is the People: SLAs, DORA and Trust at Billions of Requests a Month
By: Harry Bouras
Role / LinkedIn: AI & Cloud Engineering leader, formerly heading the API organisation of a leading, extremely global and multicultural corporation. LinkedIn: https://www.linkedin.com/in/harry-bouras/

Abstract:
Most "platform" talks are about tools. This one isn't. Heading the API organisation of a leading, extremely global and multicultural corporation, interconnecting decades of legacy systems while serving billions of requests a month and scaling across both Azure and Google Cloud, taught me a contrarian lesson the hard way: at that volume, platforms don't fail on technology, and cloud bills don't explode because of bad Terraform. They fail, or succeed, on how the organisation is set up to own, deliver and grow the thing. The org chart is the real architecture.

I'll walk through the levers that actually moved the needle, each anchored in real moments from that scale. How SLAs, framed as clarity rather than punishment, change engineer behaviour more than any dashboard. Why extending trust and time to a team up front - the most uncomfortable thing a leader can do, is precisely what raises build quality and cuts the rework you can't afford at billions of requests. What genuinely changed when we dragged a waterfall-shaped organisation toward iterative, Scrum-based delivery, and why that was about shortening feedback loops, not adding ceremonies. And how DORA metrics became our objective answer to a question every team argues about: what actually makes a successful Senior Software Engineer and what a DevOps engineer?

Then the human core, told honestly, which means told as comedy. What makes a good manager in this world, narrated through the things I was certain about and was completely wrong about. The running joke is also the real point: nearly every instinct that made me a good engineer made me a mediocre manager until I unlearned it.

Finally, a look at what we're all about to get wrong together: AI is forcing DevOps to grow up into DevSecOps - and the teams that handle it well will be the ones that already have the ownership, trust and growth culture this talk is really about.

Talk 2
Postmortem Culture: Turning Failures into Learning, Not Blame
By: Dani Yelovitch
Most engineering teams treat incidents as embarrassments to bury — a quick fix, a vague Slack message, and everyone moves on. This talk argues for a different approach: treating every failure as a gift.
The core idea is blameless postmortems — a practice pioneered at Google and Netflix where the goal isn't to find a scapegoat, but to understand how the system (people, tools, processes) allowed a failure to happen in the first place. If a human made a mistake, the real question is: why did the system make that mistake easy to make?
The talk would cover:
What a good postmortem looks like (and the common ways they go wrong)
How to run a blameless review without it becoming a blame-by-committee session
Turning action items into actual change, not a graveyard doc nobody reads
Building psychological safety so engineers report problems early instead of hiding them
The key takeaway: teams that learn from failure faster than their competitors ship better software with fewer catastrophic incidents — not because they have better engineers, but because they've built better feedback loops.