About

I believe we improve software development teams by always considering the combination of process and technology, and the way AI changes our processes makes this clearer than ever.

How I got here

I started from a background in software engineering and mathematics and even did my masters thesis on a computer-verified correctness proof of a 10-line algorithm. This methodology is not yet usable outside an academic setting, and after joining the workforce I quickly learned how to use a combination of process, reviews and tests to manage software quality. Instead of just doing the work, I always tried to find how we could optimize these processes. At Progress I improved the velocity by reducing a 12 week release cycle to a weekly release, by improving the build process, introducing feature flags and improving the agile process. At Reconcept I helped the backend and frontend software engineers come together as one team, so we had fewer bottlenecks and a more predictable velocity.

In 2025 I was asked to join Kilo, and worked on the leading edge of developing with AI, evaluating agentic coding practices and seeing the impact of AI on the development process.

What I learned along the way

Process and technology cannot be separated

At Progress I helped reduce a 12 week release cadence to a 1 week release cadence. To achieve this I asked the customers what aspect of stability the old cadence guaranteed them. Since the most prominent aspect was that changes to the software required lots of internal communication on their side, we introduced feature flags, giving them control over when the features were released to their internal users. We also improved the build process and brought the build/test time down from 4 hours to 20 minutes, so we could guarantee the main branch was deployable at every moment. Lastly we aligned the sprint length and the release cadence.

Quality Enables Speed

Managing software quality is crucial: if your quality gets too low, code gets hard to change, which means declining development speed and this hurts the business. I’ve often seen people put quality against speed/pragmatism, but if you want to go fast, you need to have quality. Writing software is cheap, maintaining software is where most of the cost is.

AI Changes how to achieve quality

Quality stays important, even when you’re working with AI. However, how we manage quality changes. Software engineering is changing because of AI. I’ve seen how software practices I’ve learned over the years as common sense don’t work anymore, but also how software quality stays important, but that it works in another way.

How I work today

Most teams are aware when they have a problem and know the symptoms, but are not always able to pinpoint the root causes. Because of that, I work embedded in the engineering team for a few weeks before proposing changes. This gives me time to see where the bottlenecks and problems are, and where AI can help. At the same time, I can already help the team by pitching in on the work.

Once I have a clear picture of the process, tech stack, and automation, I suggest improvements, including where AI could help, and once we agree, we start right away.

Why I do this work

I’ve always found it very satisfying to be able to help teams reach their full potential. And I noticed that often someone with a fresh perspective can provide a lot of benefits in a few months.