Steffen Nordnes

Across machine learning, data infrastructure, and product work, getting a demo to run is rarely the hardest part. The harder and more valuable work is turning that first version into something a team can rely on, change, and trust with real data and real users.

I try to work that way across projects, whether I am building an ML lifecycle, a backend system, or a mobile app. I like to define the scope early, agree on what success looks like, and ship work in reviewable increments instead of hiding everything until a final reveal. Before something goes live, I want a clear check that it behaves as expected. After release, real usage should inform the next iteration, with a rollback path if something breaks.

I value clear explanations and practical decisions, especially when a system has to be maintained by more than one person. That is why I treat reproducibility, automation, and observability as part of the build, not paperwork added at the end. A system you can rebuild, test, and watch is much easier to improve without guessing, and much more likely to survive beyond the demo.