Project Overview
The AI Recipe Assistant is a sophisticated Retrieval-Augmented Generation (RAG) system built to provide accurate recipe information based on specific document data. Developed with Python Django and LangChain, it serves as an intelligent bridge between culinary knowledge bases and user queries.
By leveraging vector embeddings and high-performance LLMs via the Groq API, the assistant can understand complex user requests regarding ingredients and preparation methods while strictly adhering to the provided source material (PDFs).
Key Features
- Intelligent Ingredient Querying: Users can list ingredients they have, and the agent identifies relevant recipes from the knowledge base.
- Method Extraction: Provides detailed, step-by-step cooking instructions sourced directly from PDF texts.
- Strict Data Grounding: The agent is programmed to only answer questions based on the provided PDF text, preventing hallucinations and ensuring accuracy.
- High-Speed Inference: Utilizes Groq Cloud for ultra-fast response times.
The Challenge
One of the primary challenges was ensuring the model didn't provide generic recipes from its pre-trained knowledge. By implementing a strict RAG pipeline with ChromaDB and custom prompt engineering, I ensured that every response is cited and grounded in the uploaded culinary documents.