UCBMealz
UCBMealz is a subscription-style meal planning tool I built that generates personalized meal recommendations based on dietary preferences and nutritional goals, using real data scraped from UC Berkeley's own dining hall menus. I used Selenium to scrape and structure the menu data, then built a React front end where users could set their preferences (protein, carbs, calories, dietary restrictions) and get meal suggestions generated in real time through Google's Generative AI (Gemini) API. It was as much a lesson in wrangling messy scraped data and designing good prompts as it was in building the app itself. Getting the model to consistently respect nutritional constraints while still suggesting food people actually wanted to eat took a lot of iteration.