AI Engineer M-F, Photographer on the Weekend.

I started as a data engineer, building Python pipelines, Snowflake models, and dbt transformations that turned messy, sparse data into clean client-ready marts. That work made me allergic to AI fluff. I like a good demo as much as anyone, but I care more about what happens when the system fails. We treat data pipelines that way, so why should AI pipelines be any different? It is why I built a document extractor as a deterministic, governed tool, not a prototype that looks great until it meets a real document.

Now I am building an AI design team around the same belief: a model is only as good as its input. Every project at work has a PM and documentation, and models deserve the same. Specs, artifacts, constraints, and asking "why" are what make vibe coding actually hold up. My spicy take: AI will replace us, so do not be the one getting dumber with every prompt. Think while you build.