WASTE WATCH AI

Inspiration

Food wastage is a major challenge for food service establishments, especially small hotels and messes where food preparation is often based on manual estimation and past experience. We were inspired by the idea that data such as past consumption, weather, holidays, events, and attendance could be used to make food preparation more accurate and reduce unnecessary waste.

What We Built

We developed WasteWatch AI, a data-driven food demand prediction system that estimates the quantity of food required for a particular day. The system processes historical consumption and waste data along with contextual factors and uses machine learning to predict expected demand and recommend an optimal preparation quantity.

How We Built It

We designed the system with a React.js frontend, Python and FastAPI backend, Pandas and Scikit-learn for data processing and machine learning, and a database for storing historical consumption, waste, and prediction data. Our architecture follows a continuous pipeline of data collection → processing → prediction → recommendation → feedback.

What We Learned

Through this project, we learned how real-world factors can influence demand prediction and how machine learning can be applied to solve an operational problem. We also gained experience in connecting a frontend, backend, database, and ML model into a single application.

Challenges

One of our main challenges was designing a prediction approach that could account for multiple changing factors rather than relying only on historical sales. Another challenge was designing a feedback mechanism where actual consumption and food waste can be used to improve future predictions.

Our Vision

WasteWatch AI aims to move food preparation from guesswork to data-driven decision-making. In the future, we envision integrating real-time weather, attendance, smart waste tracking, inventory management, and demand-based menu planning to create a more intelligent food-management system.