I am interested in understanding complex systems through mathematical models, computational methods, and intelligent algorithms. Whether the application concerns environmental dynamics, healthcare, or scientific data infrastructure, the underlying scientific challenge stays the same: constructing models that are mathematically rigorous, computationally reliable, and useful for decision making. Everything below follows from that one challenge.
The Centerpiece
Research Questions
I don’t organize my work around projects first. Projects exist because of these questions.
How can mathematical models better represent complex environmental systems?
Seasonal forcing, delayed effects, and incomplete measurements make environmental systems resistant to simple models. I'm interested in how much structure a compartmental approach can capture before it needs to become something more complex.
Related: Mathematical Modelling · Environmental Modelling
How can scientific computing improve the reliability of numerical simulations?
A simulation is only as trustworthy as the stability of the scheme producing it. I look at how discretization choices (finite differences, operator splitting) decide whether a result reflects the physics or just numerical artifacts.
Related: Scientific Computing · Numerical Analysis
How can artificial intelligence support scientific discovery while remaining transparent and reproducible?
Machine learning can find structure I wouldn't specify by hand, but only if it stays interpretable enough to validate. I'm interested in architectures that combine symbolic reasoning with learned components without losing that transparency.
Related: Artificial Intelligence · Machine Learning
How can data infrastructures improve reproducibility in computational research?
Most data platforms are built for business scale. I'm interested in what changes when reproducibility, not throughput, is the primary design constraint.
Related: Scientific Data Infrastructure
How can mathematical modelling contribute to healthcare decision support?
A model that informs a health decision carries a different burden of proof than one that doesn't. I'm interested in decision-support tools precise enough to be useful and honest enough to be questioned.
Related: Healthcare Analytics · Biostatistics
The Vocabulary
Research Themes
Ten perspectives on the same underlying problem, not a checklist of skills.
Applied Mathematics
Why it matters. Every model in this research program starts as a mathematical object before it becomes anything computational.
Scientific challenge. Formalizing a real system precisely enough that the mathematics constrains the model, rather than the other way around.
Current application. Compartmental ODE systems, group-theoretic structures, differential equations underlying every project below.
The language everything else here is written in.
Scientific Computing
Why it matters. A correct equation with no way to evaluate it is not yet useful.
Scientific challenge. Discretizing continuous mathematics without discretizing away the physics it describes.
Current application. Finite-difference and operator-splitting schemes for wave propagation and atmospheric flow models.
The bridge between a proof and a prediction.
Numerical Analysis
Why it matters. Simulation results are only as trustworthy as the stability of the scheme producing them.
Scientific challenge. Guaranteeing convergence and stability under the specific constraints of the problem, not just in theory.
Current application. Stability analysis for finite-difference schemes applied to strongly damped wave equations.
The quality control layer beneath scientific computing.
Mathematical Modelling
Why it matters. Complex environmental and health systems rarely have closed-form descriptions; they need models built for them.
Scientific challenge. Capturing seasonal, non-autonomous dynamics without overfitting a model to a single dataset.
Current application. Non-autonomous compartmental modelling of chlordecone environmental transfer and exposure.
Where applied mathematics meets a specific, named scientific question.
Artificial Intelligence
Why it matters. Some structure in complex systems is easier to learn from data than to derive analytically.
Scientific challenge. Keeping learned models transparent and validated rather than treating them as black boxes.
Current application. A multi-agent research framework combining symbolic reasoning with machine learning.
Another modelling framework, held to the same rigor as the mathematical ones.
Machine Learning
Why it matters. Classical statistical learning remains the most interpretable way to turn data into a decision signal.
Scientific challenge. Choosing models simple enough to explain their own reasoning to a non-specialist.
Current application. Random Forest-based risk scoring for clinical pre-triage prioritization.
Where AI research becomes a usable decision-support tool.
Healthcare Analytics
Why it matters. Health decisions carry real consequences, which makes them the highest-stakes test of whether a model can be trusted.
Scientific challenge. Building decision support that is fast and interpretable without oversimplifying the underlying risk.
Current application. Pre-triage prioritization tooling and public-health exposure modelling.
The destination this entire research program is oriented toward.
Biostatistics
Why it matters. Health and environmental data are noisy by nature; statistics is what keeps conclusions honest about that noise.
Scientific challenge. Quantifying uncertainty in exposure and outcome estimates without overstating precision.
Current application. Statistical grounding for exposure modelling and risk estimation work.
The discipline that keeps healthcare-facing models accountable.
Environmental Modelling
Why it matters. Environmental systems are where seasonal forcing, delayed effects, and human exposure intersect.
Scientific challenge. Representing rainfall-driven, seasonal contamination dynamics with a model simple enough to interpret.
Current application. Chlordecone contamination modelling in tropical ecosystems.
A direct, concrete instance of mathematical modelling applied to public health.
Scientific Data Infrastructure
Why it matters. A research result that can't be reproduced by someone else isn't finished, and reproducibility starts with the data layer.
Scientific challenge. Designing infrastructure for reproducibility first, scale second, which most analytics stacks don't optimize for.
Current application. A multi-tenant analytics platform integrating DuckDB, Apache Iceberg, dbt, and Trino.
The engineering discipline that makes every other theme here reproducible, not just correct.
Interdisciplinary Balance
Scientific Profile
Read this less as a skills inventory and more as a map of balance. Each axis depends on the others: mathematical modelling means little without the computing to test it, and computing means little without the statistical grounding to trust its output.
Methodology
Research Framework
How a question here typically becomes an answer.
- Scientific Question
- Mathematical Model
- Numerical Method
- Scientific Computing
- Data
- Machine Learning
- Validation
- Decision Support
Evidence
Research Outputs
Publications are one research output among several, alongside software, technical notes, and the projects above.
Preprints & Research Notes
Published in ResearchGate (Preprint), 2026
Preprint on compartmental modeling of human exposure to chlordecone in tropical environments.
Recommended citation: Geovany B. P., Laguerre (2026). "Modélisation de l'exposition humaine au chlordécone." ResearchGate Preprint. DOI: 10.13140/RG.2.2.36238.01607
Download Paper
Published in ResearchGate (Presentation), 2025
Presentation of the fundamental concepts of finite fields and character theory (additive characters, orthogonality, and key results).
Recommended citation: Geovany B. P., Laguerre (2025). "Exploring Finite Fields and Their Character Theory." ResearchGate Presentation. DOI: 10.13140/RG.2.2.13837.45285
Download Paper
Published in Manuscript in preparation, 2026
Work in preparation on ODE-based modeling for environmental risk and decision support.
Published in Manuscript in preparation, 2026
Work in preparation on integrating mathematical foundations into Data Science and AI training paths.
Published in Manuscript in preparation, 2026
Work in preparation on a modern analytics architecture for reproducible scientific workflows.
Commitments
Open Science
None of this work is credible to me if it can’t be checked. That means code that runs for someone other than me, methods documented well enough to be challenged, and results that don’t depend on a specific machine to reproduce. GitHub hosts the code and technical notes; Geo’s Stories hosts the less formal record of how the ideas actually developed.
- Version Control
- Open Source
- Reproducible Pipelines
- Documented Methods
- Public Technical Notes
Directions, Not Promises
Future Research
- AI for Science
- Scientific Machine Learning
- Environmental Mathematics
- Healthcare Analytics
- Computational Biology
- Decision Support Systems
- Digital Health
- Mathematical Epidemiology
These are directions I want to grow into, not results I’m claiming. The throughline that got me here (read on the Scientific Journey page) is the same one pointing toward them.