devopsdays · phrawzty · Sep 13, 2026 · Sep 9, 2026 · Sep 12, 2026 · Sep 12, 2026
@@ -0,0 +1,16 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "Terraform State Is Not Just a File"
+Type = "talk"
+Speakers = ["david-kubec"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+"Terraform state is often treated as just a file stored in a remote backend. In larger infrastructures, however, it becomes a critical part of the system: a source of truth, a synchronization point, and an operational risk.
+In this five-minute lightning talk, I will show cases where monolithic Terraform infrastructure is not the result of laziness or poor design, but of real dependencies between components. Splitting state into smaller pieces can help, but sometimes it only creates a distributed monolith with an extra orchestration layer.
+Based on this problem, I will introduce KiloLock: an open-source, self-hostable HTTP backend for Terraform/OpenTofu state, backed by PostgreSQL. In its basic mode, it works as a vanilla remote backend that a team can operate itself. Because the state is database-backed, it can also add value beyond a typical blob-based backend — for example state queries, change history, or resource-level locking.
+The talk is a technical look at why it may make sense to operate Terraform state as infrastructure in its own right, not just as a blob in a bucket."
@@ -0,0 +1,23 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "How We Reduced Costs By Migrating Across Clouds: Lessons from a Real World Platform Migration"
+Type = "talk"
+Speakers = ["gerrard-ezeugwa"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+This is the story of how we decided to move from one cloud solutions provider to another, after years of questioning our monthly cloud expenditure and responding to multiple questions from our finance department.
+
+We understood the magnitude of the task, and so we rolled up our sleeves and got to work. What began as a series of meetings with cloud providers where we had conversations on technology, architecture, support and commercial models, evolved into a proof of concept: migrating our UAT environment to a new cloud platform. 
+
+From there, the scope grew rapidly. What started as a Kubernetes migration soon expanded to object storage, networking, DNS and hosted zones, container registries, identity and access management, CI/CD pipelines, database and other supporting services. Before long, we were planning and executing production migration. This all happened while keeping the business running.
+
+Some decisions saved us weeks of effort. Others created unexpected problems for us. We learned not only the dependencies between resources, but also the sequence in which to move them to minimise disruption
+
+This talk is not about why one cloud provider is better than the other. It is a candid account of engineering decisions. It is also about the mistakes we made and how we learned from them in a real-world cloud migration effort driven by cost optimisation.
+
+Whether you are planning a cloud migration, questioning your cloud bill or simply planning a large-scale infrastructure change, you will leave with practical lessons that apply regardless of the cloud provider you use.
@@ -0,0 +1,14 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "Don't make the same mistakes when operating JVM in K8s"
+Type = "talk"
+Speakers = ["jakub-coufal"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+Lessons and fuckups learned the hard way from operating JVM based services in K8s from the perspective of Day 2 operations.
+The session is going to be a rapid-fire of pitfalls to avoid and remediations and alternatives to apply.
@@ -0,0 +1,13 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "GPUbernetes: Playing with expensive stuff in Kubernetes on AWS"
+Type = "talk"
+Speakers = ["jakub-stehlik"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+LLMs, AI agents and machine-learning models are all the rage now. Everybody uses them. Running them as Kubernetes apps in the cloud, however, is a bit tricky - at the very least, you need expensive specialized hardware (GPUs) and a mindset tweak to work around the special nature of these workloads. Let's delve into what you need to do, what tools are available, and what fun and frustrating challenges await you when you decide to run containerized GPU workloads in Kubernetes in the cloud today - might actually change tomorrow, who knows?
@@ -0,0 +1,17 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "Your website is not healthy just because humans can open it."
+Type = "talk"
+Speakers = ["nadav-harari"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+A production deployment can return 200 OK, pass uptime monitoring and still make an entire section of a website disappear from Google.
+
+Search engines do not only ask whether a page is available. They ask whether it is crawlable, indexable, renderable and associated with the correct canonical URL.
+
+In this 5-minute talk, I will show how seemingly harmless releases can introduce invisible SEO incidents through noindex directives, broken redirects, canonical changes, crawler blocking and missing content. I will then introduce a lightweight “search reliability” layer: automated pre-deployment checks, post-deployment crawling and alerts for business-critical SEO changes.
@@ -0,0 +1,13 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "Build your home AI lab with local LLMs"
+Type = "talk"
+Speakers = ["petr-hanak"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+Open-source local LLMs are a good opportunity to gain hands-on experience with AI. No extra hardware needed. Just open-source software and a few dollars are all you need to build your "home AI lab". In this short talk you'll see a real demo in action: virtual hardware provisioning, software installation, and a real use case in VS Code.
@@ -0,0 +1,13 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "AI Helped Us Respond Faster… and Debug Slower"
+Type = "talk"
+Speakers = ["saikeerthi-m"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+We brought AI into our incident response workflows to move faster—quicker debugging, smarter fixes, less operational stress. At first, it worked. AI suggested fixes, generated runbooks, and helped us respond faster than before. It felt like we had upgraded our entire DevOps workflow overnight. Until it didn’t. We started seeing fixes that looked correct but failed in production, recommendations that didn’t fully make sense, and engineers quietly losing trust in the system. Incidents weren’t just technical problems anymore—they became judgment calls between human intuition and AI suggestions. In this 5-minute Ignite talk, I’ll share what actually happened when we let AI into incident workflows—the wins, the failures, and the unexpected lesson: AI doesn’t reduce complexity, it shifts it.
@@ -0,0 +1,17 @@
++++
+Talk_date = ""
+Talk_start_time = ""
+Talk_end_time = ""
+Title = "Fugu & Containers: Handle With Care (please!)"
+Type = "talk"
+Speakers = ["szymon-izydorek"]
+youtube = ""
+slideshare = ""
+slides = ""
++++
+
+What do Fugu fish and Docker containers have in common? Well… they’re lethal if mishandled, and both require skill, patience, and a deep understanding of what you’re working with. In Japan, only licensed chefs are allowed to prepare and serve fugu. One wrong move can be fatal.
+
+Now ask yourself: is the same true for your infrastructure? Are your Docker containers properly secured and hardened? No --privileged flag? Secrets properly managed? Minimal attack surface?
+
+In this 5-minute Ignite, we’ll talk about Docker security—how to handle containers safely, avoid poisonous mistakes, and keep your systems alive and thriving.
@@ -0,0 +1,5 @@
++++
+Title = "David Kubec"
+type = "speaker"
+linktitle = "david-kubec"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Gerrard Ezeugwa"
+type = "speaker"
+linktitle = "gerrard-ezeugwa"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Jakub Coufal"
+type = "speaker"
+linktitle = "jakub-coufal"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Jakub Stehlík"
+type = "speaker"
+linktitle = "jakub-stehlik"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Nadav Harari"
+type = "speaker"
+linktitle = "nadav-harari"
++++
@@ -4,8 +4,8 @@ type = "speaker"
 linktitle = "oleksii-mudryk"
 image = "oleksii-mudryk.jpg"
 linkedin = "https://linkedin.com/in/moudrick"
-github = "https://github.com/moudrick"
-twitter = "https://x.com/moudrick"
+github = "moudrick"
+twitter = "moudrick"
 website = "https://moudrick.net/"
 +++
 

@@ -0,0 +1,5 @@
++++
+Title = "Petr Hanak"
+type = "speaker"
+linktitle = "petr-hanak"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Saikeerthi M"
+type = "speaker"
+linktitle = "saikeerthi-m"
++++
@@ -0,0 +1,5 @@
++++
+Title = "Szymon Izydorek"
+type = "speaker"
+linktitle = "szymon-izydorek"
++++
@@ -81,8 +81,12 @@ organizer_email: "prague@devopsdays.org" # Put your organizer email address here
 sponsors:
   - id: labyrinthlabs
     level: gold
-  - id: minimus
+  - id: echo
     level: gold
+  - id: nebius
+    level: lanyard
+  - id: sky
+    level: silver
   - id: hackihub
     level: community
   - id: alicetechnologies
@@ -91,6 +95,8 @@ sponsors:
     level: community
   - id: art4people
     level: community
+  - id: quanti
+    level: community
 
 sponsors_accepted: "yes" # Whether you want "Become a XXX Sponsor!" link
 
@@ -136,7 +142,7 @@ program:
     date: 2026-10-05
     start_time: "10:30"
     end_time: "10:55"
-  - title: ""
+  - title: "gerrard-ezeugwa"
     type: talk
     date: 2026-10-05
     start_time: "10:55"
@@ -192,11 +198,17 @@ program:
     end_time: "?"
 
 ignites:
-  - title: ""
+  - title: "jakub-coufal"
+    date: 2026-10-05
+  - title: "saikeerthi-m"
     date: 2026-10-05
-  - title: ""
+  - title: "petr-hanak"
     date: 2026-10-05
-  - title: ""
+  - title: "david-kubec"
     date: 2026-10-05
-  - title: ""
+  - title: "nadav-harari"
     date: 2026-10-05
+  - title: "jakub-stehlik"
+    date: 2026-10-05 
+  - title: "szymon-izydorek"
+    date: 2026-10-05