A rundown of what I've built, the stack behind each one, and how it went.
Started this as STAR. It's an AI academic assistant: you tell it what you need in plain language and it turns that into lesson plans, task schedules, and opportunities worth looking at. I built it so students could stay on top of their own goals instead of juggling five different apps.
1st place ($1,000), Tacos & Tech-ila Audience Pitch Competition.
Visually impaired people in the El Paso/Juárez metro had no navigation tool built for them, not for this border, not in both languages. So we built one: a smart cane prototype on an Arduino R4 WiFi, paired with a web app. You say where you're going, in English or Spanish, and it guides you there with real-time GPS, AI voice input, and Google Maps routing.
1st place, BorderHack '26 — 32 teams, 70+ competitors, built in about 24 hours.
I started Black Box because I lived the problem: years doing theater and film, practicing alone, with no real way to know if I was getting better. It's a web app for actors to rehearse on their own, scene analysis, script practice, recorded performance review, plus a computer vision feature that tracks your face and pose live while you practice.
3rd place at Miner Tank and 3rd place at the P21 VMS Pitch Challenge. Now the flagship product under Lyra Solutions.
Two summers at Bloomberg's Tech Lab on campus. In 2025 I built a full-stack news feed with React, Flask, Redis, and Docker. In 2026 I worked through a RabbitMQ producer-consumer system: queues, routing, acknowledgements, and debugging message flow across containers.
Bloomberg Tech Lab on Campus, 2025 and 2026.
Built this at the Borderland AI Hackathon: an assistant that helps K-5 teachers find book collections by grade, subject, and theme instead of digging through catalogs by hand. Used LangChain so it actually retrieves and recommends real book data instead of just generating text.
1st place, Immersive Learning track, plus the Innovator Award.
Built for a one-day Datathon: a classification model that predicts high-activity trading days from real S&P 500 data. I went through the whole pipeline myself, feature engineering, training four different models, evaluating them, and defending every decision in a live presentation.
1st place, ACM-W UTEP & ColorStack at UTEP Datathon.