A mix of shipped products, ML research, and weekend experiments — spanning native apps, transformers, and recommender systems.
Ten years of real, official PLCB shelf prices for ~31,000 wines, with gradient-boosted models forecasting each wine's chance of a price rise or cut and a hedonic fair-price valuation (80% of log-price variance explained on held-out wines).
A hybrid movie recommender — SVD collaborative filtering, genre similarity, and a Bayesian quality prior trained on the MovieLens dataset — wrapped in a React Native app styled like a movie theater lobby.
PhotoNester is a native iOS app that uses Core ML and CLIP vision embeddings to classify and cluster photos into albums based on visual similarity, all processed securely on-device.
A live companion dashboard for an ESPN Fantasy Basketball league: today's top performers and matchup scoreboard, season-long point leaders with per-player trend sparklines, and daily stat snapshots backfilled into Postgres via a scheduled Netlify Function.
MIT capstone project for the Applied Data Science and AI program: built and compared CNN architectures, including a VGG16 transfer-learning model, to classify parasitized vs. uninfected blood cell images for faster, more accessible malaria diagnosis.
Built the marketing site and client portal for a Jamaica-based personal shopping and international shipping business, with a Firestore-backed public reviews feature and Google Analytics with consent management.
Marketing site for my own AI/ML consulting studio: machine learning strategy, custom model development, automation pipelines, and MLOps for businesses, with an RFP intake flow and lead-magnet signup.
A couple of apps I'm actively shipping right now — not quite live yet, but you can preview each on its own site.
An iOS mood-journal app pairing an Expo/React Native frontend with a FastAPI emotion-analysis backend, so you can track how you're feeling over time. Currently in App Store review.