💡 Inspiration We wanted to create an AI fitness assistant that combines trustworthy health info with the power of language models to give users smart, real-time guidance.
⚙️ What it does It’s a fitness chatbot that answers health and nutrition queries using a custom knowledge base, semantic search, and OpenAI’s GPT for intelligent responses.
🛠️ How we built it Collected data from 3 fitness sources Split text into 500-token chunks Generated embeddings using llama-mini-v6 Stored vectors in Pinecone Used LangChain RAG with OpenAI API Built a simple web UI using Flask + Bootstrap
🚧 Challenges we ran into Managing token limits during prompt construction Balancing speed and accuracy in retrieval Setting up Pinecone and embedding pipelines smoothly
🏆 Accomplishments Successfully deployed a working RAG chatbot Integrated multiple tools into a seamless pipeline Stored and searched over 12,000 text chunks effectively
📚 What we learned How RAG pipelines work in real-world apps Hands-on with Pinecone, LangChain, and OpenAI APIs Designing scalable vector-based search systems
🔮 What’s next Add user authentication and history Build personalized fitness plans Improve UI and add voice/chat app integrations
A DEMO VIDEO IS ADDED TO THE GITHUB REPO UNDER THE NAME 'Fitness Bot'. MAKE SURE TO DOWNLOAD AND LOOK AT IT!
Log in or sign up for Devpost to join the conversation.