Estrada Maldonado, J. O.
EN ES

Jesús Osvaldo Estrada Maldonado

I measure how algorithmic systems and environmental exposure reshape human wellbeing, and I write the software that makes those measurements hold up.

Economist and research assistant at the University of Guadalajara. Two published studies spanning environmental health economics and the psychology of algorithmic platforms, and a third on machine learning for outcome prediction currently in press.

Open to research & data engineering roles
ORCID
0009-0000-9002-8282
Affiliation
University of Guadalajara
Dept. of Society and Economics
Based in
Guadalajara, Jalisco, Mexico
Focus
Data science · Applied economics · Computational social science
Output
2 published · 1 in press
Stack
Python · R · SQL · PostgreSQL
Languages
Spanish (native) · English (professional)

Depressive experience, ages 15–29

73.2%

of young people in the Guadalajara metropolitan area (23.7% of them frequently).

Encuesta Juventudes Guadalajara 2024
Read the study

Annual productivity forgone

€1.18bn

lost to pollution-linked cognitive impairment in Małopolskie, over 96% of the region's total pollution cost.

EEA & Eurostat, 2013–2018
Read the study

PM2.5 against WHO guideline

5–7.7×

the recommended average concentration, sustained across six years in the same region.

European Environment Agency
Method note

Research

Two threads that turn out to be one question: what happens to people when a system, an atmosphere, a recommendation algorithm, exerts pressure nobody chose and few can measure.

2 published · 1 in press

Identidades violentadas y riesgos psicológicos en la era algorítmica

Ma Teresa Prieto Quezada · Jesús Osvaldo Estrada Maldonado · José Claudio Carrillo Navarro · Leonardo Javier León Aguilar

Introduces algorithmically mediated violence (aggression that platforms do not merely carry but actively produce, amplify and modulate) and documents its psychological correlates among young people in Guadalajara. Proposes a five-level ecosystemic intervention framework in place of individualised, pharmacological or techno-phobic responses.

Facebook70.2%
Instagram62.6%
TikTok61.7%
Platform use among surveyed youth. Facebook's persistence contradicts global narratives of its youth obsolescence, evidence of locally specific cultural dynamics.

Economic impact of air pollution on health and productivity in Małopolskie Region, Poland

Jesús Osvaldo Estrada Maldonado — sole author

Six years of particulate exposure in southern Poland set against health and labour outcomes. The headline result is a reframing: the dominant economic cost of air pollution is not respiratory mortality but the quiet, cumulative loss of cognitive productivity, a channel most air quality policy does not price at all.

Respiratory deaths+23%
Hospitalisations+26%
Cognitive-loss share96%
Health outcomes in 2018 against 2013, and the share of total economic cost attributable to cognitive productivity loss rather than direct health expenditure.

How the work gets made

Every study above runs through the same pipeline. I build all of it, there is no separate person who writes the code.

End to end

  1. Sources

    Surveys, EEA monitoring, Eurostat registries, press coverage

    SQL · scraping · APIs

  2. Clean & join

    Schema design, ETL, validation against source

    Python · pandas · PostgreSQL

  3. Model

    Estimation, machine learning, simulation

    statsmodels · scikit-learn · R

  4. Validate

    Repeated runs and cross-validation before anything is claimed

    100+ seeds · significance testing

  5. Publish

    Figures, tables, the written argument

    LaTeX · Quarto · matplotlib

Version controlled end to end · Git

Every stage re-runnable from raw source

The standard I hold it to: someone else should be able to run the analysis a year from now and get the same number.

Data engineering

Building datasets that did not previously exist in usable form, from survey instruments, public registries and scraped text.

Python · SQL · PostgreSQL · DuckDB · pandas · schema design · ETL

Statistical modelling

Estimation and inference on small, awkward, real-world samples, where the honest answer is often "not distinguishable from nothing".

statsmodels · R · tidyverse · exposure–response modelling · significance testing

Machine learning

Transfer learning, embedding fusion and gradient boosting, with evaluation protocols built to survive the variance of small datasets.

scikit-learn · transformer embeddings · gradient boosting · cross-validation

Communication

Turning an estimate into a figure, a paragraph and a policy implication a non-technical reader can act on.

LaTeX · Quarto · matplotlib · Spanish (native) · English (professional)

Background

Economics gave me the questions. Programming is how I answer them.

2022 — present

  • Research assistant

    University of Guadalajara. Dataset construction, statistical processing, literature review and preparation of results for publication.

  • BA in Economics

    University of Guadalajara, Department of Society and Economics, Zapopan, Jalisco.

Contact

Research collaborations, data work, or an opening you think fits.

Replies within a few days

jesus.osvaldo.estrada.maldonado@gmail.com

Based in Guadalajara, Jalisco. Available for remote work, in Spanish or English.