Currently: AI Safety Research Engineer at Equistamp, contracting with Redwood Research on the LinuxArena project (co-author, accepted at NeurIPS 2026)
6 years ML Engineering in industry + 1 year building network simulations at CERN
Co-director of Argentinian AI Safety: general management, principally governance and comms
I curate Spotify playlists that everyone loves
Nerd out on breathing techniques and cognitive performance somewhere between pranayama and Huberman's sleep-and-focus hacks. Ask me anything about optimizing your neurology
Favorite movie: Pulp Fiction. Favorite book: Rayuela by Julio Cortázar
Contracting with Redwood Research on the LinuxArena project, building evaluation infrastructure for AI safety. Co-author of the LinuxArena paper, accepted at NeurIPS 2026.
Developed discrete‑event simulation of data acquisition system (4TB/s) in C++ to design upgrades.
Built containerized environment for cross‑platform simulation replicability.
ML Engineer · MercadoLibre
Aug 2017 — Sep 2023
Fraud Prevention (Jul 2019 — Sep 2023)
Built a computer vision model for handwritten digit detection (MNIST transfer) and an NLP model for personal data detection (400% impact vs. baseline).
Deployed a gradient boosted trees model to detect undelivered-product scammers.
Built distributed Spark pipelines on EMR for terabyte‑scale data processing.
Recommendations (Aug 2017 — Jul 2019)
Trained and deployed collaborative filtering models (item‑to‑item) for supplementary and complementary recommendations.
Built data pipelines processing millions of events daily.
