🍽️ FoodGPT — AI-Powered Restaurant Recommendation System

AI-powered restaurant discovery system that understands natural language queries and returns personalized recommendations using a Retrieval-Augmented Generation (RAG) pipeline.

---

🧠 How It Works

---

✨ Features

• 🔍 Natural language search (e.g. "cheap chicken under ₹300")
• ⚡ Semantic search using FAISS + sentence-transformers
• 🤖 LLM-powered responses using Ollama (Phi / LLaMA3)
• 🔄 Real-time streaming responses (SSE)
• 💬 ChatGPT-style UI (Next.js)
• 🍽️ Structured restaurant cards (name, rating, cuisine, location)
• 🧠 Prompt grounding to prevent hallucinations

---

🏗️ Tech Stack

Backend

• FastAPI
• FAISS
• Sentence Transformers
• Ollama (Local LLM)

Frontend

• Next.js (App Router)
• Tailwind CSS

---

📂 Project Structure

foodgpt/
│── backend/
│   │── main.py
│   │── rag_engine.py
│   │── embeddings/
│   │── data/
│   │── build_index.py
│
│── frontend/
│   │── app/

---

⚙️ Setup Locally

1️⃣ Clone the repository

git clone https://github.com/your-username/foodgpt.git
cd foodgpt

---

2️⃣ Backend Setup

cd backend
python3 -m venv .venv
source .venv/bin/activate

pip install -r requirements.txt

---

3️⃣ Install Ollama (LLM)

curl -fsSL https://ollama.com/install.sh | sh

Run:

ollama pull mistral
ollama serve

---

4️⃣ Build FAISS Index (one-time)

cd ..
python build_index.py

---

5️⃣ Start Backend

cd backend
fastapi dev main.py

Backend runs on:

http://127.0.0.1:8000

---

6️⃣ Frontend Setup

cd frontend
npm install
npm run dev

Frontend runs on:

http://localhost:3000

---

🧠 Architecture

User Query → Embedding → FAISS Search → Context Injection → LLM → Streaming Response

---

📸 Screenshots

Chat Interface

Recommendations Panel

---

🚧 Future Improvements

• 🗺️ Map integration
• ⭐ Rating filters
• ❤️ Favorites system
• 📊 Sorting & personalization

---

💡 Key Learnings

• Built a complete RAG pipeline from scratch
• Implemented real-time streaming with SSE
• Controlled LLM hallucination via prompt grounding
• Designed full-stack AI product with real UX

---

⭐ Star this repo if you like it!