Portfolio.
Applications and projects in production
MLM Studio
Shiny app for complete Mixed Linear Model analysis without writing code. Covers the full pipeline: import, descriptives, longitudinal profiles, individual OLS estimates, mixed model with lme4, and nested model comparison. Includes automatic natural language interpretation, ICC with Cicchetti classification and lambda shrinkage factor.
TimeSight
Professional time series analysis platform with modular architecture (bs4Dash). Implements walk-forward CV, Interval Score, residuals ANOVA, ARIMA/ETS/TBATS models, Bayesian components with bsts, and actionable automatic diagnostics. Designed for non-programmers with theoretical tooltips on every element.
La Tienda — Price Intelligence
Wholesale price intelligence system for Antioquia retail merchants. Implements real-time Normal-Normal conjugate Bayesian updating, IQR anomaly detection, exponential smoothing, STL decomposition and Mann-Kendall test. Data from Central Mayorista, DANE-SIPSA, Banrep and WFP. Proactive alerts and 15-day prediction.
SymFonos — Visual Chaos of Your Music
Generative WebGL music visualizer that turns real-time audio (mic or file) into 3D animations driven by chaotic physics equations: spring pendulum and double pendulum integrated with RK4 for high precision. Dynamic color in LCH space, custom GLSL shaders (background, masses, chromatic aberration) with bloom, live Lyapunov exponent, chaos mode with random impulses, zen mode, and 4 visual presets (Kandinsky Pulse, Vignelli Grid, Neon Filament, Particle Swarm) inspired by Reas, Ikeda, Klingemann, Kwok, Akten, Rist and Zobel. Every session is unique and unrepeatable.
@RIOS — Simulation & Optimization
Monte Carlo simulation and stochastic optimization engine in Shiny with advanced modular architecture. Allows defining random variables (Normal, LogNormal, Triangular, PERT), decision variables and constraints in natural R language. Genetic algorithm optimization (GA) with common samples and parallelization with future/furrr.
Statistical Software Engineering
Open source manual for taking R/Python statistical models to production. Documents architecture, DataOps, MLOps and data governance: statistical software principles, model management with MLflow, feature stores, monitoring, compliance checklist, CI/CD, rollback, Shiny/React best practices, and a maturity roadmap for data teams.