ingrai - 🍳 Ingredient Tracker Chatbot

demo_video.mp4

A Streamlit-based chatbot with persistent memory that uses tools to keep track of your ingredients and preferences.

• Food preferences and dietary restrictions
• Ingredients you have available at home

The application uses a PostgreSQL database to remember chat history, preferences, and ingredients.

Technology Stack

• Frontend: Streamlit
• AI Framework: LangChain and LangGraph
• Language Model: Qwen3 14B (running locally)
• Database: PostgreSQL for persistent storage
• Architecture: Agent-based system with memory management

Project Structure

• app.py: Streamlit application and user interface
• agent.py: Implementation of the conversational agent with LangGraph
• memory.py: Memory management and database persistence layer
• models.py: Pydantic models for data structures
• prompts.py: System messages and instructions for the language model

Setup and Installation

Prerequisites

• Python 3.9+
• PostgreSQL
• Ollama (for running the Qwen model locally)

Installation Steps

1.Clone the repository:

git clone git@github.com:ccrvos/ingrai.git
cd ingrai

2.Install dependencies:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

3.Set up PostgreSQL:

# Make sure PostgreSQL is running on port 5442
# Example using Docker:
docker run --name postgres-ingrai -e POSTGRES_PASSWORD=postgres -p 5442:5432 -d postgres

4.Install Ollama: Follow Ollama installation instructions

5.Run the Ollama server:

ollama serve

6.Open new terminal and run the Ollama model:

ollama run qwen3:14b

7.Run the application:

streamlit run app.py