Computational social science and AI learning
I study how social and economic differences shape who uses, trusts and learns with AI. I program fluently in Python and R, and I work with survey data, social networks and text.
Before my MSc I spent eight years running IT and support at a school in Brazil, four of them also teaching programming. I hold degrees in computer science, systems analysis and development, science and technology, and law.
MSc Computational Social Science, Linköping University. Swedish Institute Scholar. Research intern at the Institute for Analytical Sociology.
Linköping University. Working on a method that places survey-based distance matrices on the surface of a sphere and tests whether it preserves them better than flat PCA and MDS maps.
Independent, survey-weighted analysis of 26,404 respondents across the EU27 using Special Eurobarometer 554.
Designed and built an open-source Python pipeline that collects and codes university guidance on generative AI in five Nordic countries.
Supported the running of the 8th EUSN, hosted by the Institute for Analytical Sociology in Norrköping in August.
For the programming course I designed and taught for students from low-income communities in Brazil.
Awarded to international students with strong academic records and leadership potential. Funds my MSc at Linköping University.
Flat PCA and MDS maps distort distances between respondents, parties or countries. I test whether placing the same distance matrix on the surface of a sphere preserves it better, comparing Kruskal stress across both approaches.
Across the EU, how comfortable people feel with AI at work depends far more on their own digital skills than on education or age. People who have already met AI at work are more accepting, and financial precarity lowers acceptance in a way digital skills do not explain.
A reproducible pipeline that collects and codes official university guidance on generative AI. It scores each policy on two separate axes, restriction and pedagogical support, instead of a single strict versus permissive scale.
What a survey of more than 26,000 Europeans says about who accepts AI at work, and why that turns AI literacy into a question of inequality.
Most talk about university AI rules sorts them into strict or permissive. Reading the documents themselves, I think that misses how they actually work.