About
How people understand —
and what I build to find out.
I get curious about how people understand, and I keep building tools to test those theories. Conversational AI by trade. Madrid by choice. Trilingual (English, Spanish, Hebrew).
Four facets
01
Product
I lead from user value — clear hypotheses, sharp scoping, and a habit of cutting scope down rather than up. Most of the leverage in product is in deciding what not to build.
02
Strategy
Build-vs-buy, evaluation frameworks, the connective tissue between research and roadmap. I write the memos that lock down a decision before it gets eroded by month two.
03
Builder
I ship side projects top-to-bottom, from idea to deployed URL, because keeping my hands on the keys keeps my taste calibrated.
04
Rigorous data person
BigQuery, dbt, Python. CUPED, variance reduction, power analysis. The instinct that "we can probably tell if this works" is a starting point, not a conclusion.
Background
Trained in cognitive science (MSc, Brain & Cognitive Sciences — EEG complexity measures for disorders of consciousness). Since then, product work centered on a production conversational-AI system: evaluation frameworks, experimentation infrastructure, and the daily craft of steering model behavior with evidence. The side projects are where the four facets stay honest — each one shipped end to end, from idea to deployed URL. Based in Madrid; works in English, Spanish, and Hebrew. Always happy to talk evidence-based product, evaluation design, or how people understand things.