Yammi

Inspiration

I was inspired to create Yammi during the "MLH Hack Your Portfolio" hackathon because I wanted to address a common issue faced by many home cooks and food enthusiasts. Often, we find ourselves missing a key ingredient while preparing a dish, and finding a suitable substitute without compromising the flavor can be challenging. I wanted to leverage AI/ML to provide a solution that makes cooking more accessible and less stressful.

What it does

Yammi is an AI/ML web-based application that helps users find suitable substitute ingredients for their recipes. By simply entering the ingredient they want to substitute and the recipe, Yammi suggests alternatives that can be used without compromising the dish's flavor and texture. This ensures that users can continue cooking their favorite meals even if they don't have all the required ingredients on hand.

How I built it

I built Yammi using NextJS for the frontend and integrated TensorFlow.js to power the AI/ML model. The model is trained on a dataset of common ingredients and their substitutes, allowing it to make accurate predictions. The design of the website is inspired by the Kawaii style to make it visually appealing and user-friendly. I also applied my UI/UX skills to enhance the overall user experience.

Challenges I ran into

One of the main challenges I faced was creating an accurate and comprehensive dataset of ingredients and their substitutes. Ensuring the AI model could understand and provide relevant suggestions was another hurdle. Additionally, integrating TensorFlow.js with NextJS and optimizing the model for quick and accurate predictions required significant effort.

Accomplishments that I'm proud of

I'm proud of successfully creating an AI/ML model that provides reliable ingredient substitutes. The Kawaii-inspired design of the website turned out beautifully, making it not only functional but also delightful to use. I'm also proud of the positive feedback received during the hackathon, which validated the usefulness and appeal of Yammi.

What I learned

Throughout this project, I learned a lot about integrating AI/ML models into web applications using TensorFlow.js. I also gained deeper insights into data preprocessing and model training. Moreover, I enhanced my skills in UI/UX design, particularly in creating visually engaging and user-friendly interfaces.

What's next for Yammi

The next step for Yammi is to expand the ingredient database and improve the accuracy of the AI model. I plan to add more features, such as providing nutritional information for the substitutes and enabling users to save their favorite substitutions.

Additionally, I aim to make Yammi a mobile app to reach a broader audience and make it even more accessible using Flutter and Firebase database with Tenserflow library.

Built With

• nextjs
• postman
• tenserflow