About Me

I’m a software engineer with a background in energy trading systems. Between 2022 and late 2025, I worked at Norlys Energy Trading, joining when the company was still building its infrastructure from scratch and staying through a period where it scaled from around 80 people to over 300. That growth meant working across a lot of ground: forecast and live market data, asset operation and remoting, reconciliation, bid generation, communication with grid operators, and the enterprise trading systems where those pieces converged. Building through that kind of expansion, where the business was still figuring out what it needed while we were building it, teaches something about how complexity compounds when the boundaries between systems are not thought through clearly enough.

In October 2025, I joined Voltique Energy as a senior software developer. The company is built on the premise that energy trading should be modular: businesses starting out should be able to acquire only the capabilities they need rather than staffing entire departments from the outset. Here, system design and architecture are central to the work, and autonomous AI agents started intersecting with that work in ways that felt worth paying attention to. The question that emerged was not whether agents could automate what was already well understood, but how far their reasoning could extend into problems that had not been fully specified yet.

That intersection is what this blog explores.

What keeps pulling at me is how to specify what done looks like clearly enough that an agent understands what you actually want. Writing a specification that an agent can execute against without constant clarification turns out to require the same precision that makes shared understanding possible in human teams, and I have been spending a lot of time thinking through why that parallel holds and what it implies for how we structure technical work more broadly.

I started writing about this because I needed to work through the thinking myself. The posts here are grounded in work I have actually done, not frameworks assembled in the abstract. I am still learning, genuinely open to different approaches and to being wrong about what I think I know, and most interested in thinking through these questions with others who are asking them too.