About Me
My Education
I graduated from Lincoln Preparatory Academy in 2024. I'm now a Sophomore at Arizona State University in Computer Systems Engineering!
My Passions
I am deeply passionate about exploring cutting-edge technology, including AI trustworthiness, sustainable computing, and AI Optimization.
Accomplishments
Flinn Finalist in the 2024 Cohort and a National Hispanic Scholar.
Proudly a part of ARC Lab with Professor Zhou at ASU!
My Hobbies
I love rocks, I love rock climbing, I do fencing, coding goofy passion projects, and cooking (as best I can) with friends!
My Projects
Developing LLM Bias Metric @ ARC Lab w/ FURI | Current
I'm working with Professor Ben Zhou and PhD Student Xiao Ye, looking into how pretraining can affect model performance and trying to create a metric that can determine a model's trustworthiness! A tough challenge for sure, but so very, very incredibly interesting!
Machine Learning from Scratch in Game Engines | Current
This has been one of my favorite projects! In a game called Rec Room, which uses a node-based programming language with zero libraries, I created an MLP and a CNN completely from scratch. Rec Room has some fun computational restrictions, like only 16k operations at a time, and no nested lists!
Local Jarvis - My Assistant | 2025
We all know Tony Stark's AI butler that helps him around the house. While I'm not a billionare, I did spend the past year creating a custom LLM with memory formation, long-term memory retrieval, self-developing personality, and the ability to control certain aspects of my computer, all hosted on my local computer!
1800's Trained AI Model | 2025
I had a blast learning to build an LLM from scratch. From the attention blocks, transformers, tokenizers, and data-loaders, I learned a ton. I decided to train a model on around 1.7m tokens of 1800's text. It turned out decently well, just a bit too much memorizing.
KodaCo Auditor | 2024
I built a medical auditor for KodaCo. It uses OpenAI's API to read through audit logs of patients and ensure proper patient care. It uses carefully crafted prompts developed alongside medical professionals to ensure the highest levels of patient safety.
WizardTyper | 2024
A fun project that had me working as a full stack developer. I created the website from scratch, including all the frontend and backend functionality. It uses Python, JavaScript, HTML, CSS, and a framework through Flask, SQLAlchemy, and Bootstrap.
