A novel Meta-Reinforcement Learning (Meta-RL) framework for adaptive scheduler selection in offline, non-preemptive mixed-criticality scheduling on varying-speed processors[cite: 13].
A time-indexed mixed-integer programming framework incorporating chance constraints and a grid-based linearized Chernoff-bound approximation to address inventory estimation uncertainty in post-wildfire salvage logging.
Fusing machine learning for fault detection with stochastic optimization to orchestrate robust, resource-aware maintenance schedules under uncertainty.
In this column, I discuss the limitations of purely predictive analytics and explain why organizations must adopt Operations Research to transition from looking at past data to making optimal decisions.
In this interview (starting at 40:45), we discuss the practical impact of data science in Chile, focusing on prescriptive analytics in healthcare, traditional industries, and territorial management.
Our team participated in OPTIMA 2025 with three contributions across healthcare operations, observatory log analytics, and industrial maintenance. Presentations covered robust routing for home hospitalization, decision oriented relevance ranking for system logs at Paranal Observatory, and sensor based degradation modeling in bottling lines, highlighting our interdisciplinary approach to operations research, machine learning, and decision support.
An argument for moving from prediction to prescription where high-quality data, explicit uncertainty, and decision governance can turn models into actions that withstand scrutiny.
Conversation with PLANEO Magazine on how data science can support just and sustainable cities, from descriptive, predictive, and prescriptive analytics to governance, equity, and collaboration.
Interview featured in *El Mercurio - Chile Tecnológico* highlighting how data science tools tripled oncology patient care in a pilot project at Hospital de La Florida, developed in collaboration with Universidad Católica.