Marcel Pons Cloquells

My default has always been to get something real into production as fast as possible, then improve from there. That rhythm shaped how I think about both data engineering and data science: define what's enough to be useful, ship it, watch how it behaves, sharpen it.

I've built most of my career in small data teams of two or three people. That context shaped how I work: I own the full stack across data engineering and modeling, which means a faster path from question to answer, and answers that hold up in production.

I work at Cloud and Big Data scale, across the full data stack. In the last year I've added Claude Code, and it's genuinely changed how I work. I spend more time on the parts that require judgment, and I'm constantly testing in practice how far AI can go in data work: what it can replace, what it can augment, and where human judgment still matters more than people think.

I care about clean abstractions, reproducible environments, and systems that explain themselves. The right system multiplies everything built on top of it. Getting that foundation right is always worth the time.

When I'm not wrangling pipelines, I'm reading, writing about data strategy, or exploring how technology shapes the way we make sense of things.