About
Machine learning engineering shaped by actuarial thinking.
I build machine learning systems for decisions where uncertainty has a real cost: credit, pricing, underwriting, and insurance.
My path into ML began with mathematical engineering and actuarial science. That foundation still shapes the way I work: start with the decision and its constraints, quantify risk, then design the data and operational systems that make a model dependable in practice.
Today, my work connects statistical modeling with production engineering. I am particularly interested in model monitoring, feature platforms, reproducible pipelines, and the interface between ML outputs and the people responsible for business and risk decisions.
SELECTED EXPERTISE
From model design to dependable operation.
I work where model quality, data reliability, and business constraints meet.
01
Production ML & MLOps
Deployment workflows, model registries, monitoring, batch scoring, APIs, and feature stores built for maintainable operation.
02
Credit, risk & pricing
Credit scoring, churn, underwriting, actuarial pricing, reserves, and portfolio analytics grounded in financial constraints.
03
Data & feature platforms
Reproducible pipelines and feature systems across Snowflake, Databricks, Airflow, Feast, MLflow, AWS, Docker, and Kubernetes.
SELECTED IMPACT
Outcomes over output.
1 week → 20 min
Model-monitoring metric retrieval
5h → 45 min
Feature processing across 20M transactions
−22%
Default rate through credit scoring
FEATURED WRITING
Notes on models, methods, and systems.
01 · EN · R WORKFLOWS
Create a CV with R Markdown and Google Sheets
A reproducible workflow for building and maintaining a polished CV from structured data.
02 · ES · INSURANCE
Introducción al seguro
An introduction to the actuarial ideas, history, and risk-transfer mechanics behind insurance.
03 · ES · SIMULATION
Juego de la Vida de Conway
Cellular automata, emergent behavior, and a practical Python implementation of Conway’s Game of Life.
PROFILE
An actuarial foundation. An engineering practice.
My background in mathematical engineering and actuarial science shapes how I approach ML: define the decision, make uncertainty visible, and build the operating system around the model—not only the model itself.
How I work
01
Begin with the decision
A useful model has a clear user, action, risk boundary, and definition of success before the first training run.
02
Design for operation
Data contracts, reproducibility, monitoring, and ownership are part of the product—not implementation details left for later.
03
Communicate uncertainty
Technical quality includes making assumptions, trade-offs, failure modes, and business consequences understandable.
Education
- Diploma in Actuarial Science, Pontificia Universidad Católica de Chile · 2019
- Mathematical Engineering, Universidad Central del Ecuador · 2016
Focus
Production ML · MLOps · Credit risk · Pricing · Underwriting · Actuarial analytics · Feature platforms