Final-year Systems Engineering and Industrial Engineering student at Universidad de Ibagué, Colombia. I build web and mobile apps and data pipelines with Python, JavaScript and SQL.
- Turquestra. A PWA for tour operators: conflict-free bookings, customers and preventive equipment maintenance. I led development in a team of three; one of the three winners of the Pijaotech startup competition (Festech 2026). In use by its first operator. React, Spring Boot, PostgreSQL.
- Zociva. Fertilizer recommendations for coffee growers, my undergraduate thesis in a research group. It combines IDEAM weather-station data, CHIRPS satellite rainfall and OCR of soil lab reports; a linear-programming model picks the lowest-cost fertilizer mix. Offline-first PWA for use on the farm. Python, FastAPI, React, PostgreSQL. Private code: ongoing research.
- Sports forecasting data platform. ETL pipelines with Polars and Pydantic into PostgreSQL, a gradient-boosting ensemble (XGBoost, CatBoost, LightGBM) and more than 3,000 automated tests.
- Field software for an engineering company. An offline app for technical site surveys with photos, sketches and customer signature, office review, PDF reports and quotes. TypeScript, Fastify, PostgreSQL. Private under a confidentiality agreement.
- ISO 20022 instant payments. A simulator of Colombia's Bre-B scheme, written in Java from Banco de la República's public technical specification: funds reservation and settlement, timeouts with unknown outcomes, daily reconciliation, mTLS. The repo goes public when its first phase is done.
- A harness for AI coding agents. Guardrails that block known agent cheats before they run, an action log the agent can't edit, and a cost per accepted change.
- Spec and tests first; every behavior traces to a requirement.
- Numbers I measured, not adjectives.
- Security in the pipeline, not after the release.