I’m a Virgo. And while I trust data far more than horoscopes, this constellation felt like a good way to connect everything I do.
LLM-based systems built end to end — architecture, integration, evaluation and monitoring — with a focus on latency, reliability and business metrics.
A conversational agent deployed in production on WhatsApp. A cognitive architecture built on LLMs, connected to the channel through the messaging provider Jelou, with an end-to-end flow and handoff to human advisors.
Conversational and automation agents operating on real processes, beyond the prototype.
A conversational agent deployed in production on WhatsApp. A cognitive architecture built on LLMs, connected to the channel through the messaging provider Jelou, with an end-to-end flow and handoff to human advisors.
An AI agent that automatically generates commercial communications in HTML, automating a marketing process that used to be manual.
Natural language and speech processing applied to conversations, interactions and documents.
A tool deployed on the collections phone channel: automatic classification of non-payment reasons, analysis of advisors’ soft skills and speech recommendations.
Sentiment analysis of customer interactions to prioritise experience improvements in collections.
Automated processing of legal documentation for collateral recovery workflows.
Research on generative AI applied to learning, built on systems of my own.
AIAI 2026 — 15th Workshop on Mining Humanistic Data (MHDW)
Ordóñez Olazábal, J. S., et al. (2026). GenAI-Powered Feedback for Video-Based Learning. AIAI 2026 — 15th Workshop on Mining Humanistic Data (MHDW). Springer.
A full-stack application that gives teachers feedback on video courses, built from data capture to user interface. It is the foundation of my AIAI 2026 publication.
Over seven years in digital banking (BCP / Credicorp): product, strategy and analytics before AI.
−38%operational calls, through a differentiated call-centre service initiative
+37%sales effectiveness on prioritised customers, by improving the lead-prioritisation model
Universidad de Buenos Aires
AI specialisation: deep learning, computer vision and NLP; IoT and application development.
University of London
Upper second-class honours.
Universidad del Pacífico
Graduated second in my class.
University teaching in natural language processing and artificial intelligence, and thesis supervision.
Artificial intelligence is also built through conversation, shared experience and learning from other people. Here I gather the talks, conferences and workshops I take part in.
What does it take for an artificial intelligence initiative to stop being an experiment and become a solution used at scale?
In this conference I’ll explore the path from the first trials to putting AI solutions into production: the technical challenges, the strategic decisions and the conditions needed to drive real adoption.
Because building a model is only the beginning. The real challenge is making AI work outside the lab.
Universidad San Ignacio de Loyola (USIL)For collaborations, research, teaching or conversations about applied AI.