-
First the question and the decision
Why do rural schools barely use the digital library? What was the Central Bank discussing during the crisis? Before opening R I define the question and, above all, which decision depends on the answer. The design, the sources and the precision needed follow from that.
- Research design
- Surveys
- Interviews
-
Then data you can trust
Validating the 2024 Census taught me that most of the work is cleaning, joining and documenting. Explicit rules, versioned code someone else can rerun and written assumptions.
- R
- SQL
- Python
- Git
-
Analysis looks for differences that matter
Statistical models, network analysis or automated text classification when the corpus is large. The method is chosen by the question and reported with its limits, including null results.
- Statistical models
- GPT-4o
- Network analysis
- STM
-
And it ends in something the institution uses
A report, a dashboard or a reporting pipeline the team can rerun without me. At SLEP Santa Rosa that means schools get their indicators up to date and anyone can audit where each figure comes from.
- Power BI
- Quarto
- GitHub Actions
- Claude Code
Professional experience
Work, projects and consulting
Experience across the public sector, private companies, academic research and consulting. What I did on each project, with which tools and what came out of it.
How I work
From a question to a decision
Every dot is a person, a listing or a document. My job is to arrange them until they say something useful.
720 loose dots
CV
Five years, four worlds
Official statistics, think tanks, strategy consulting and public education. Each one taught me to talk to a different audience.