juanse hevia.
about

hi. I'm juanse.

engineer, founder, and Fulbright scholar — I work between research and deployment, technical and non-technical, startups and the public sector. across LLMs, computer vision, and classical methods, the constant is the same: pick the right tool, then build it.

moments

here, in pictures.

✺ a few places this work has taken me.
portrait of juanse
01portrait — 2024
presenting in Buenos Aires
02buenos aires — speaking
MIT policy hackathon presentation
03MIT — policy hackathon
Rice University campus
04rice — fulbright year
at the Einstein memorial in Washington DC
05DC — einstein memorial
Washington DC trip
06DC — capitol hill
founder identity

professional experience.

2025—

founder · in stealth

building tools that help non-technical people put AI to work

more after launch.

2022—2024

ML engineer · head of data management

Memorable AI · acquired by Reddit

shipped applied computer vision under client pressure. designed the data infrastructure and processing pipeline that supported 10× storage growth and AdExchanger-recognized work.

2021—

AI, innovation & government coordinator

Universidad Austral

academic programming and applied materials that help public-sector professionals connect AI foundations with governance, policy, and institutional use cases.

rigor

academic experience.

2023—2025

MS data science · Fulbright scholar

Rice University

graduate work in applied ML and a NASA capstone on lunar crater detection — record precision, presented to NASA engineers in Houston JSC.

CVPR · PMLR · JPSP

peer-reviewed publications

three venues, three angles on applied ML

Modular Memorability (CVPR 2023), Accessible AI Tutors (PMLR 2025), ML-SPEAK (Journal of Personality and Social Psychology 2025).

2017—2021

BSc · teaching assistant

Universidad Torcuato Di Tella

undergraduate foundations in economics and CS. early research and teaching that set the through line: rigor in service of usability.

from startup AI systems to NASA research to public-sector programs — the through line is simple: I move between worlds, judge which tool actually fits, and build the thing that ships.