Introduction

Welcome to the Indian Food Companion documentation. This guide will teach you how to build a complete AI-powered food recommendation application from scratch.

What You'll Build

By the end of this guide, you'll have built a fully functional chat application that:

• Understands natural language - "I want something spicy for breakfast"
• Provides personalized recommendations - Based on dietary preferences and allergies
• Streams responses in real-time - Just like ChatGPT
• Uses your own curated data - Not just relying on the AI's training data

Architecture Overview

┌─────────────────────────────────────────────────────────────────┐
│                         User Interface                          │
│                    (React + shadcn/ui)                          │
└─────────────────────────┬───────────────────────────────────────┘
                          │
                          ▼
┌─────────────────────────────────────────────────────────────────┐
│                      FastAPI Backend                            │
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────────────┐  │
│  │   Chat API   │  │ Preferences  │  │    MCP Server        │  │
│  │   (SSE)      │  │     API      │  │   (AI Tool Access)   │  │
│  └──────┬───────┘  └──────────────┘  └──────────────────────┘  │
│         │                                                       │
│         ▼                                                       │
│  ┌──────────────────────────────────────────────────────────┐  │
│  │                  RAG Pipeline                             │  │
│  │  1. Query → 2. Retrieve from ChromaDB → 3. Generate      │  │
│  └──────────────────────────────────────────────────────────┘  │
└─────────────────────────┬───────────────────────────────────────┘
                          │
          ┌───────────────┼───────────────┐
          ▼               ▼               ▼
    ┌──────────┐   ┌──────────┐   ┌──────────────┐
    │ ChromaDB │   │ Postgres │   │  OpenAI API  │
    │ (Vectors)│   │  (Users) │   │  (LLM)       │
    └──────────┘   └──────────┘   └──────────────┘

Prerequisites

Before starting, you should be comfortable with:

• React - Component-based UI development
• REST APIs - HTTP methods, JSON, request/response patterns
• Docker - Basic containerization concepts
• Python - Basic syntax (we'll teach you FastAPI)

You don't need prior experience with:

• Large Language Models (LLMs)
• Vector databases
• Embeddings
• RAG (Retrieval-Augmented Generation)
• MCP (Model Context Protocol)

We'll explain all of these concepts as we go.

Tech Stack

Component	Technology	Why We Use It
Frontend	React + Vite	Fast development, modern tooling
UI Components	shadcn/ui	Beautiful, accessible, customizable
Styling	Tailwind CSS	Utility-first, rapid prototyping
Backend	FastAPI	Async Python, automatic API docs
Vector DB	ChromaDB	Simple, embeddable, perfect for learning
Database	PostgreSQL	Reliable, feature-rich
LLM	OpenAI	Best-in-class, easy to use
Tools	MCP	Standardized AI tool protocol

How This Guide is Organized

1. Getting Started - Set up your development environment
2. Core Concepts - Understand LLMs, RAG, embeddings, and MCP
3. Backend Development - Build the FastAPI server step by step
4. Frontend Development - Create the React chat interface
5. Deployment - Containerize and run with Docker

Let's begin!