SERVESMART

Project Logo

Inspiration

ServeSmart is inspired by the need to reduce food waste and help people. Many fresh products are thrown away daily, while some struggle to afford food. We aim to connect businesses with communities to save resources and promote fairness.

What it does

ServeSmart is an innovative platform designed for restaurants, markets, bakeries, and similar businesses to announce their fresh but unsold products at the end of the day. The platform aims to reduce food waste while enabling individuals in need to access these products at more affordable prices. By preventing the unnecessary disposal of resources, ServeSmart promotes a sustainable consumption model that benefits both businesses and the community. Our AI-integrated system helps sellers announce their products more effectively by leveraging text generation and time-series models to optimize listings and predict demand. With its user-friendly interface and mission-driven approach, ServeSmart not only addresses the pressing issue of food waste but also fosters a greener future.

How we built it

To build our project we used Python, Streamlit (for our interface). Our project contains text-generation and time-series models for better user experience. Text-generation : We've added a powerful text generation feature in Sell Product tab of our application. At first, we tried to fine-tune Llama and Gemma models, and actually we did it. But there was an option called Gemini. We did the tests and decided to use gemini-1.5-flash. So we decided to use Gemini API. After hitting the "Submit Product" button, we're sending an API request to Gemini (gemini-1.5-flash) and improve the title-description of the meal. Time-series : Our application uses a time series model to analyze product pricing data entered by sellers. The model considers factors like holidays and time of day to predict customer traffic. Built with machine learning frameworks, it uses historical sales data and traffic patterns for accurate predictions. If traffic exceeds a set threshold, a backend algorithm dynamically applies discounts to boost sales and reduce waste, ensuring both customer and seller satisfaction.

Challenges We Faced And Accomplishments We’re Proud Of

Integrating advanced AI features and ensuring real-time performance pushed us to think creatively. Balancing simplicity with functionality helped us make the platform more user-friendly and impactful. We’re proud to have created a platform that helps reduce food waste while supporting businesses and communities. Adding AI for smart predictions and price adjustments was a big success for us.

What We Learned

We had the chance to use our AI knowledge and improve our skills on platforms like Google Colab. This project showed us that every year, a lot of food is wasted, while many people still struggle to find good food. Seeing that technology can help solve these problems gives us hope for the future.

What's Next for SERVESMART

Our next step is to turn our prototype into a real website and mobile app. After launching, we will keep improving it based on user feedback. We hope our project will inspire others to take action for sustainability.