Explore the power of container security with Kirill Shirinkin as he guides you through enhancing your CI pipeline with Clair, a tool that identifies vulnerabilities in container images. Learn to automate security scans using Clair to safeguard your deployments: https://mkdev.me/posts/scan-container-images-with-clair-v4

What can we do better to better protect people from fraud? More details in our conversation with Katarina Pranjić from LexisNexis Risk Solutions: https://mkdev.me/posts/preventing-digital-fraud-with-katarina-pranji-from-lexisnexis-risk-solutions-55

#mkdevWeeklyHighlight
Apparently Google Cloud has a decent native tracing solution, integration of which this article by Punit Sethi explores: https://punits.dev/blog/building-observability-with-google-cloud-services/

How did AI affect fraud? There are many ways, some of which are predictable, and some a very new. Full discussion with Katarina Pranjić from LexisNexis Risk Solutions in episode 55 of DevOps Accents: https://mkdev.me/posts/preventing-digital-fraud-with-katarina-pranji-from-lexisnexis-risk-solutions-55

Horror story time! In a land not so far away from mkdev's headquarters in Munich one insurance company almost had to cancel their IT migration. Because their legacy system was so old, it couldn’t generate the test cases needed to validate their new setup, turning a critical IT migration into a nightmare. 👻

That’s when our Head of Data & AI, Paul Larsen, stepped in. 🧙‍♂️

With his leadership, we built an AI-driven automation solution that:
✔️ Transformed dense insurance documents into a reference pricing engine, automatically converting specs into Python code.
✔️ Extracted test cases using LLMs so nothing slipped through the cracks.
✔️ Built a “delta-explainer” that traced every discrepancy back to its source.

Weeks of manual work? Gone.
Instead, they had an automated, AI-powered process that worked in days saving time, improving accuracy, and future-proofing their operations.

Still dealing with nightmares of outdated IT slowing you down? Let’s fix that! Contact us at paul@mkdev.me for GenAI wizardry.

Is Kubernetes starting to feel less like a powerful tool and more like a complexity monster? 🤔 In the new episode of the DevOps Accents podcast we dove into a conversation with Gerrit Schumann from mogenius, and it sparked a lot of thoughts. We asked some tough questions:

❓ Has the cloud-native landscape become too intricate?
❓ Are we placing unrealistic expectations on developers to be Kubernetes experts?
❓ Is there a better way to manage this complexity and empower teams?

Short answer -- The need for simplification is real. We discussed the rise of self-service platforms and the importance of abstracting away unnecessary complexity.
It's clear that while Kubernetes is incredibly powerful, we need to find ways to make it more accessible and manageable. We need to focus on enabling developers to focus on what they do best: building great software. While many of you are experiencing Kubernetes fatigue, we also urge you to take a closer look at mogenius given they are about to release a V2 that will bring a powerful new Kubernetes Manager for development teams. More details are in the episode, you don't want to miss it!

https://mkdev.me/posts/kubernetes-has-gotten-too-complicated-with-gerrit-schumann-from-mogenius-56

Explore lightweight, secure alternatives to Docker for building container images on Kubernetes, as Kirill Shirinkin demonstrates in this article. Learn to build with Buildah, analyze with Dive, and manage registries like AWS ECR using Skopeo: https://mkdev.me/posts/buildah-dive-skopeo-3-container-tools-for-building-images-on-kubernetes-cluster-with-gitlab-ci

In the 63rd mkdev dispatch Leo draws your attention to the speed with which the artificial intelligence market is developing and the regulations that start to rise because of that. Check it out and subscribe for more: https://mkdev.me/posts/regulate-impossible-to-let-loose-63

#mkdevWeeklyHighlight
Seventh yearly report from Catchpoint on the state site reliability engineering: https://www.catchpoint.com/asset/2025-sre-report

There is a catch up game most companies play with their infrastructure automation. More on it in episode 56 of DevOps Accents with Gerrit Schumann from mogenius: https://mkdev.me/posts/kubernetes-has-gotten-too-complicated-with-gerrit-schumann-from-mogenius-56

Explore the intricate balance of tools and team dynamics in DevOps with insights from Gerrit Schumann in episode 56 of DevOps Accents. Join us for a practical discussion on how companies can streamline operations and empower developers to achieve more with less cognitive overload: https://www.youtube.com/watch?v=3sF5YK8VIzI

Unravel the true nature of Kubernetes with Kirill Shirinkin! Is it a complex tech network or the new standard for app runtimes? Discover how K3s powers CI/CD jobs efficiently: https://mkdev.me/posts/kubernetes-is-just-an-api-let-s-see-it-on-k3s-cluster-running-on-aws-spot-fleet

Gerrit Schumann from mogenius suggests investing in strong policies, documentation, and infrastructure automation to efficiently scale operations without constantly increasing team size and training burdens. More details in our conversation with him in episode 56 of DevOps Accents: https://mkdev.me/posts/kubernetes-has-gotten-too-complicated-with-gerrit-schumann-from-mogenius-56

#mkdevWeeklyHighlight
Railway.com got irritated with Google Cloud enough (“o despite multi-million dollar annual spend, we get about as much support from them as you would spending $100”) to build its own data center in just 9 months - the project that they detail in this blog post: https://blog.railway.com/p/data-center-build-part-one

Gerrit Schumann says companies want control and flexibility, but real-world needs make infrastructure complex. DevOps teams are stuck fixing issues instead of automating, though this should change. More about it in the full episode of DevOps Accents with him: https://mkdev.me/posts/kubernetes-has-gotten-too-complicated-with-gerrit-schumann-from-mogenius-56

At mkdev, we've learned something important:
AI doesn’t magically fix coding headaches—it needs smart engineering.

We put this to the test with a tedious task: refactoring legacy data warehouse code.

We fed legacy and ideal code samples into a GenAI tool, expecting quick results. Instead, we got broken outputs and failed tests. So, we rolled up our sleeves and built something smarter.

Instead of relying solely on LLMs, we designed an automated workflow that:

🤖🧹 Iterated until refactored code passed all tests.
🤖🧹 Handled functional differences, especially randomness, via a semi-automated process.
🤖🧹 Turned a manual, error-prone chore into something developers actually appreciated.

The secret? AI projects aren’t just AI—they’re smart engineering + AI.

The payoff?
✅ Reliable results.
✅ Happier developers.
✅ Faster development.

Your team deserves better than manual refactoring. We can help. Let’s talk: paul@mkdev.me or book a call 👉 https://l.mkdev.me/callconsultant

In episode 57 of DevOps Accents Alessandro Ceni, Lorenzo Cardini and Pio Scielzo from Teticum join Leo to discuss the new possibilities of AI agents for e-commerce. From identifying the problem to choosing the right tools to the issues you might face implementing your new solution, they take you through a whole journey of their startup: https://mkdev.me/posts/e-commerce-new-dirty-secret-ai-agents-with-teticum-57

In this article, Kirill Shirinkin will share his default work environment setup. You will learn how to install productivity-oriented distributions of Zsh and Vim, named OhMyZsh and Janus respectively: https://mkdev.me/posts/ohmyzsh-and-janus-set-up-productive-work-environment-in-10-minutes-or-less

In the 64th mkdev dispatch Kirill talks about CloudFlare being an underrated cloud provider that has innovatively evolved from a CDN to offering user-friendly, serverless features, uniquely positioning itself between hyperscalers and PaaS. Subscribe to get new dispatches every other week: https://mkdev.me/posts/cloudflare-redefines-what-cloud-native-is-64

#mkdevWeeklyHighlight
An overview of formal and semi formal methods AWS uses to build and extend their services in the most reliable and proven way. At hyper scale, traditional approaches that suit most companies don’t fit anymore, as AWS learned almost a decade ago: https://queue.acm.org/detail.cfm?ref=rss&id=3712057

What can creators of AI tools say to AI critics? More on productive use of AI in your product in episode 57 of DevOps Accents with Alessandro Ceni, Lorenzo Cardini and Pio Scielzo from Teticum: https://mkdev.me/posts/e-commerce-new-dirty-secret-ai-agents-with-teticum-57