This paper presents Vittles, an innovative artificial intelligence-based recipe recommendation system designed to address the dual challenges of food waste reduction and personalized meal planning. While existing recipe recommendation systems primarily focus on ingredient matching or user preferences, they often fail to consider sustainability and dietary restrictions simultaneously. We propose a novel dual-model approach that integrates natural language processing (NLP) and cosine similarity analysis to provide contextually relevant recipe suggestions based on available ingredients, with particular emphasis on utilizing leftover items. The system was trained on a comprehensive dataset of 6653 recipes from Archana's Kitchen, encompassing diverse cuisines and dietary requirements. Our methodology incorporates three key components: Comparing Ingredients (Cosine Similarity): We have a way to measure how similar two ingredients are based on their characteristics. Think of it like figuring out if apples are more like oranges or more like potatoes. Ranking Recipes (Weighted Algorithm): When showing recipes, we consider three main things: how many of the ingredients match what you want, how much other people like the recipe (based on ratings), and how popular it is overall. This helps us show you the best options first. Suggesting Swaps (Ingredient Substitution): If you’re missing an ingredient or don’t like something, we can suggest good alternatives. For example, if you don’t have butter, we might suggest using margarine instead. These findings indicate that Vittles offers a significant advancement in sustainable cooking practices while maintaining high user satisfaction through personalized recipe recommendations. This research contributes to the growing field of AI-driven sustainability solutions and provides a framework for future development in intelligent food management systems.