Skip to content
View nanounanue's full-sized avatar
💭
😡
💭
😡

Block or report nanounanue

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
nanounanue/README.md

Adolfo De Unánue

Machine learning and operations research for public policy and national-scale infrastructure.

Independent fractional CTO and AI consultant · Cofradía de Ciencia de Datos e IA

Writing, research and the full record live at unanue.mx. Open-source work starts with triage-pg.


What I work on

  • Temporal machine learning for prioritizing and targeting public programs, validated as of the date each decision was made.
  • Operations research for port terminals, hospital supply chains and electoral networks.
  • Taking models to production in public institutions, with governance, explainability and monitoring.

Guiding principle: objective functions encode political choices. A model that allocates scarce capacity decides who waits, and that decision should be stated and open to argument.

Open source

Project Description
triage-pg · MIT · DOI Temporal machine learning for public policy inside PostgreSQL: cohorts, labels, features, temporal cross-validation and model governance written as SQL. A fork of the University of Chicago's Triage.
featurizer · Apache-2.0 · DOI Deep Feature Synthesis compiled to SQL. The database computes each feature as of a date, so the feature matrix is point-in-time correct.

Both projects live in the Cofradía de Ciencia de Datos e IA organization. Earlier, at the University of Chicago's Center for Data Science and Public Policy, I co-developed Triage, an open-source framework for building and auditing models.

Background

  • 2025–present · Independent: fractional CTO and AI consultant, ML advisory.
  • 2020–2026 · Tecnológico de Monterrey, School of Government: Director of Research and Faculty (2020–2025); founded the Data Science and AI Center in 2023 and headed it until October 2026.
  • 2016–2022 · DSaPP, University of Chicago, and DSSG at Carnegie Mellon and the Alan Turing Institute: twenty projects across six countries.
  • 2013–2020 · ITAM: director of the MSc in Data Science, Mexico's first, and founding director of the Data Science Center.
  • 2008–2017 · Co-founder and CTO of data ventures: Dialetheia, OPI Analytics (Endeavor 2014) and Datank.
  • PhD in theoretical physics, UNAM.

Most of this work uses partners' sensitive data and lives in private repositories. Anonymized case studies are at unanue.mx/cases.

Elsewhere

unanue.mx · Research · Talks · CV · ORCID · LinkedIn · Substack

Pinned Loading

  1. ccd-ia/featurizer ccd-ia/featurizer Public

    Automated feature engineering for temporal data using PostgreSQL — Deep Feature Synthesis (DFS) compiled to pure SQL

    Python 2

  2. ccd-ia/triage-pg ccd-ia/triage-pg Public

    PostgreSQL-native, simplified fork of DSSG triage for temporal ML

    Python 1