The short version, then the longer one.

• Ship the smallest useful thing, then iterate. Both Tetherlab and Daredate started as scrappy MVPs that earned the right to keep existing.

• Read the source. Whether it is a paper, a framework, or someone else's code. I try to understand what is actually happening before reaching for an abstraction.

• Stay close to the user. For Daredate I ran user interviews, wrote the copy, did the marketing. For VeraConnect I sat with the marketing team and watched them use the pipeline I built.

• Write things down. Plans, decisions, post-mortems. This portfolio site itself exists because I am bad at keeping things in my head.

I am increasingly interested in the seam between agentic AI and product. Specifically, the boring, load-bearing infrastructure that makes agent systems trustworthy enough to put in front of real users. Tools like Tetherlab are my current attempt to contribute to that layer.

Outside of code: I keep moving, currently in the Netherlands after growing up in Romania and a year in Pennsylvania. I read more than is reasonable, write occasionally, and care about clean energy and democratic processes.

Four working principles I try to hold to. They are not aspirational — they are how I actually operate when I am working with you.