NutriMate

Inspiration

The inspiration for NutriMate came from the growing need for personalized nutrition and meal planning in today’s fast-paced world. We wanted to create a solution that not only suggests recipes tailored to individual dietary preferences and cuisines but also provides detailed nutritional insights for better health and well-being.

What It Does

NutriMate:

Suggests Recipes: Based on user prompts, it provides personalized recipe recommendations tailored to dietary preferences, food habits, and cuisines.
Nutritional Insights: Displays the nutritional content of recipes and ingredients to help users stay informed about their meals.
Allergy Consideration: Ensures recommendations exclude ingredients that users are allergic to, prioritizing their health and safety.

How We Built It

Database: We used Snowflake to create and manage tables, storing recipes and ingredient data as the core database for the app.
Search and Retrieval: Integrated Cortex Search to retrieve relevant data efficiently from the database.
AI Integration: Utilized Mistral-LLM for natural language processing. We implemented the classify_text function to categorize prompts into three categories: recipe, ingredients, and ingredients by name.
PDF Generation: Used the fpdf library to allow users to download recipes in PDF format.
Frontend: Developed the user interface using Streamlit, customizing the UI with explicit HTML adjustments.
Deployment: Deployed the app on Streamlit Community Cloud, ensuring seamless access.

Challenges We Ran Into

Configuring Snowflake and integrating it with the app posed some initial difficulties.
Fine-tuning the classify_text function to accurately identify prompt categories was complex.
Adjusting the UI to meet user expectations required careful design and HTML customization.
Download Buttons: Disappearing after a certain interval.
Categorizing food items according to their nutritional facts.
Running the Streamlit app in the Snowflake environment.
Finding the right dataset.
Comparing two food items based on their nutritional facts prompted an error since the program initially lacked this feature.

Accomplishments That We're Proud Of

Successfully integrating multiple technologies like Snowflake, Cortex Search, and Mistral-LLM into a cohesive application.
Creating a user-friendly app that simplifies recipe and ingredient exploration.
Enabling PDF downloads, enhancing usability and practicality for users.
Deploying the app effectively on Streamlit Community Cloud.

What We Learned

How to leverage Snowflake for database management and its integration with a frontend app.
The power of AI tools like Mistral-LLM in text classification and enhancing user interactions.
Streamlit's capabilities for creating interactive, visually appealing apps.
Overcoming deployment challenges to provide a smooth user experience.

What's Next for NutriMate

Expanding the database to include more recipes and ingredients with detailed nutritional information.
Introducing personalized meal plans and dietary suggestions using advanced AI models.
Enhancing the app's UI/UX for a more engaging experience.
Adding features like voice-based search and real-time cooking assistance.
Exploring integrations with fitness trackers to align nutrition with fitness goals.