Vehicle route optimization (Feeding San Diego)
Web app for Feeding San Diego to optimize delivery vehicle routes to 400k+ people each month
Everything I've built across AI/ML Engineering, Software Development, and Data Science.
Web app for Feeding San Diego to optimize delivery vehicle routes to 400k+ people each month
Desktop application for the San Diego Zoo to label, verify, and run classification ML models on terabytes of wildlife audio data.
Won first place in a hackathon by building a platform that spawns 100 parallel AI agents to model market reactions to news headlines and help predict whether stock prices move up or down.
Developed a hybrid recommender system for Twitch streamers using implicit-feedback ALS matrix factorization with popularity-based re-ranking, achieving 4× better precision and recall than a popularity-only baseline.
Trained and deployed an SVM model on Arduino to classify banana ripeness and flag overripe bananas for large industrial warehouse operations.
Websites for campus groups including Triton Web Dev, PERMIAS SDIA, and Triton Minecraft.
Interactive dashboard and data visualization exploring food waste patterns across the United States.
End-to-end data science report on American food nutrition: data cleaning, exploratory analysis, modeling, and communication of findings.
Won first place for best Edge AI/ML in a hackathon with a fully local and offline tool (deployed on an Arduino UNO Q) that identifies local species using an image classification CNN model, and feeds that into a voiced AI agent to deliver an interactive, park-ranger-style tour guide of each species and its ecosystem.
Regression analysis of factors influencing NVDA's stock price before and after the AI boom. Concluded that NVIDIA has shifted from being influenced by broad economic conditions (VIX, interest rates) to being driven primarily by its own financial results (EPS).