Academia
Research & Academia
Urban economics, causal
inference and GeoAI
Research Statement
My research develops computational methods to extract reliable causal evidence from
large-scale administrative and operational data. The underlying problem is methodological
rather than domain-specific: when observational data are massive, spatially structured and
drawn from administrative (not experimental) processes, conventional estimators return
results that look precise and are systematically wrong.
The doctoral thesis demonstrated this empirically at scale: a standard estimator overstated
an effect by a factor of three once benchmarked against correctly constructed controls, and I
built the computational infrastructure needed to get it right over more than 100 million
records.
This situates the work at the intersection of applied artificial intelligence, machine
learning, computational statistics and data engineering, across urban systems, transport,
logistics and infrastructure, environments where data are abundant, experiments are infeasible
and decisions carry consequences. The underlying methodology, however, is domain-agnostic: the
same identification discipline applies wherever administrative or operational microdata inform
high-stakes decisions.
Research Lines
- Reliable causal inference at scale: generalising the thesis's computational
and statistical framework into a causal-identification methodology for administrative
microdata.
- AI and machine learning as complements to causal identification: using
artificial intelligence for counterfactual construction and heterogeneity discovery while
preserving identification discipline that predictive performance alone cannot provide.
- Intelligent systems for administrative decision support: applied AI and
automated-reasoning architectures for structured decision support over large administrative
datasets.
- Computational methods for urban logistics and freight systems: extending
published work on logistics-agent behaviour towards data-driven models of last-mile network
design and micro-hub location.
- Continuous monitoring systems over public records: low-cost, reproducible
early-warning infrastructure to detect anomalous activation patterns at census resolution.
- Spatial machine learning: prediction and inference under spatial dependence,
where standard cross-validation and independence assumptions fail.
Explore Doctoral Research & Econometric Specifications →
Methodological & Technical Contributions
Doctoral research (2023–2026) built an end-to-end pipeline integrating five heterogeneous public
data sources (a national cadastre exceeding 100 million construction records, 2.96 million
property-transaction forms and daily/monthly financial series) into a unified analytical panel
of 2.49 million cells spanning 100,718 urban blocks, 16 semesters and three regions.
Data Engineering
Columnar, out-of-core processing (DuckDB, pandas) with geospatial integration (geopandas,
libpysal), implemented in Python.
Causal & Computational Statistics
Heterogeneity-robust difference-in-differences (Callaway–Sant'Anna) against never-treated
units, event-study decomposition, spatial-autocorrelation diagnostics (Moran's I, LISA), and a
custom implementation of the Rambachan–Roth partial-identification sensitivity analysis.
Reproducibility Engineering
Treats reproducibility as an engineering discipline: assertion-based validation that halts
execution on a semantic failure, declared seeds for every stochastic procedure, isolated
extraction layers for third-party audit, and synthetic-data modes enabling end-to-end
verification without access to restricted sources.
Key methodological finding: restricting the
comparison to genuinely never-treated units attenuated the conventional two-way fixed-effects
estimator by 71% (0.273 → 0.079), evidence that most of the conventional coefficient reflected
counterfactual composition rather than a treatment effect. Corroborated independently through
sensitivity analysis and validated across twelve robustness checks, including 100-permutation
spatial placebos and cluster bootstrap with 499 replications.
Reproducible Artifacts & Data
The doctoral pipeline was published as a public executable notebook incorporating
assertion-based validation, declared random seeds, documented and isolated extraction layers, and
a synthetic-data mode that reproduces the full panel structure for end-to-end verification without
access to restricted records. Classification logic was independently validated against 200
manually reviewed records (96.5% agreement). No sensitive or restricted microdata are published.
Teaching & Thesis Supervision
University teaching since 2016 across nine Chilean institutions, including Pontificia
Universidad Católica de Chile, Universidad Andrés Bello, Universidad Mayor, Universidad
de las Américas, Universidad de Santiago de Chile, Universidad Diego Portales, Universidad de los
Andes, Universidad Finis Terrae and Universidad Viña del Mar. Currently part-time faculty at the
Faculty of Engineering, Universidad Finis Terrae (Operations Management, Business Strategy,
Logistics, Innovation Workshop).
- Urban Economics & Project Evaluation: urban and regional project
management (graduate), project formulation and appraisal, urban economics, macroeconomics, land
markets and valuation, real estate finance and investment, real estate project development
- Data Science, Optimisation & Systems: mathematical optimisation,
modelling and simulation, optimal resource location, systems theory, operations management,
spatial econometrics
- Transport, Infrastructure & Logistics: logistics management, road
design, road management I–II, urban road projects, transport technologies
- Innovation & Technology Management: R&D&I project
administration, technology transfer projects, business strategy
Teaching assistant at Pontificia Universidad Católica de Chile
(2017–2022), including Optimal Resource Location (Doctoral), Infrastructure Management (Master's)
and Microeconomics (undergraduate, course conducted in English).
Thesis Supervision
Supervised nine undergraduate theses (2020–2024) in Civil Engineering, Civil Engineering in
Civil Works and Construction Engineering at Universidad Andrés Bello, Universidad de los Andes
and Universidad Mayor, across four lines: operations, logistics and supply chain; investment
appraisal and applied economics; quantitative infrastructure evaluation; and spatial analytics,
GeoAI and applied data science. Examples: López, K. (2024), Cyclologistics: a sustainable
last-mile delivery strategy, Universidad Andrés Bello; Quezada, C. (2024), Microhub
design for freight transport serving San Miguel, Universidad de los Andes; Román, C.
(2022), Optimal siting of a concrete-treatment facility, Universidad Andrés Bello
(repository).
Available to supervise undergraduate/graduate theses and to serve on doctoral and master's
committees.
Education
Ph.D. in Smart Industry
Pontificia Universidad Católica de Valparaíso
2023 – 2026
Summa Cum Laude, final grade 6.9/7.0. Advisors: Álvaro Peña Fritz, Ph.D.;
Mitsuyoshi Fukushi, Ph.D.
M.Sc. in Engineering Sciences, Transport and Logistics Engineering area
Pontificia Universidad Católica de Chile
2019 – 2026
Thesis approved in 2026: Análisis de las preferencias de localización
y segmentación de agentes logísticos en la Región Metropolitana de Santiago, Chile, on
logistics sprawl in Santiago (1985–2020) and location choice by type of logistics facility.
Advisors: Ricardo Giesen, Ph.D.; Ricardo Hurtubia, Ph.D.
M.Sc. in Smart Industry
Pontificia Universidad Católica de Valparaíso
2023 – 2024
Master of Construction
Pontificia Universidad Católica de Chile
2016 – 2018
Professional Degree in Industrial Engineering
Universidad Mayor
2011 – 2017
Certifications
Urban Logistics Summer School, Universiteit Antwerpen, Belgium (2021) · Diploma in Public
Policy, PUC (2016) · Certified Urban Appraiser, ASATCH (2014) · Diploma in Economics, PUC (2013)
· Diploma in Urban Planning and Integrated Management, PUC (2012)
Research Affiliations & Funding
Research assistant, FONDECYT (ANID), Pontificia Universidad Católica de Chile (2018–2021):
Fondecyt Regular 1180605, Indicadores de Percepción de la Forma Urbana y su Impacto en las
Decisiones de Movilidad, Localización y Desarrollo Inmobiliario; Fondecyt Regular 117049,
Análisis de Alta Resolución de Operaciones de Flotas en Entrega a Domicilio; Fondecyt
Iniciación 11160031, Evaluación y Mejora de la Eficiencia Energética en Agua Potable y
Tratamiento de Aguas Servidas.
Research affiliations: CIUDHAD, Universidad Andrés Bello (Adjunct
Researcher, Aug 2025 – Jan 2026); Faculty of Architecture, Universidad del Desarrollo (Researcher,
Jan–Feb 2025); ISCI, Institute for Complex Engineering Systems (Doctoral Student Researcher,
2019–2021); CIGIDEN, Centre for Integrated Disaster Risk Management (Doctoral Student Researcher,
2018–2019).
A proposed independent research agenda seeks external funding through
competitive national and international instruments, building on this FONDECYT experience and on
existing ties with researchers at PUCV, Pontificia Universidad Católica de Chile and Universidad
Andrés Bello.
Professional & Institutional Service
Doctoral degree in Smart Industry awarded Summa Cum Laude, first graduating cohort of the PUCV
programme (2026). Council Member and Subcommittee Coordinator, Infrastructure for Sustainable
Mobility, Regional Government of Valparaíso (2015–2018): led technical-policy roundtables across
the 12 municipalities of Greater Valparaíso, contributing to a policy proposal presented to
Chile's Ministries of Public Works, Transport and Housing (MOP, MTT, MINVU).
