FoodNav

This is our initial MVP for AI and Food Insecurity Case Competition by Robert H. Smith School of Business, University of Maryland. We have developed an AI based solution which uses a combination of advanced NLU, LLMs(Generative Ressponses) and a multilingual agent to help people find food assistance resources in the DMV area.

Our data is from Capital Area Food Bank, which is a non-profit organization that provides food assistance to people in need in the DMV area.

Installation for Local Development

Server Setup:

•   cd server
    python -m venv venv
    source venv/bin/activate  # On Windows use `venv\Scripts\activate`
    pip install -r requirements.txt

• Set up environment variables in a .env file. Follow the .env.example file for reference.

• Start the server using:

    uvicorn app.main:app --reload

Database Setup:

• Run the seed.py script to populate the database with initial data:

    python app/seed.py

Client Setup:

• Navigate to the client directory and install dependencies:

    cd client
    pnpm install

• Start the client application:

    pnpm run dev

Agent Setup:

• Our agent is not available in this repository. If you need access to it feel free to contact me.

Features:

• Call Support: You can dial the number provided to talk with the agent directly and get responses in real-time.

• Call Companion: You can you can use the call companion feature to get assistance while on a call with the agent. This feature provides real-time support and guidance during the call.

• Multilingual Support: The agent can understand and respond in english and spanish. More languages wil be added in the future.

• Generative Responses: The agent can generate responses based on the user's input and the data from the Capital Area Food Bank.

• Advanced NLU: The agent uses advanced natural language understanding techniques to understand the user's intent and provide relevant responses.

• Data-Driven: The agent uses data from the Capital Area Food Bank to provide accurate and relevant information to the user.

• Web Interface: The agent is accessible through a web interface, allowing users to interact with it easily.

• User-Friendly Interface: The client application provides a user-friendly interface for users to interact with the agent.

Future Improvements:

• Add accurate location collection for the Spanish agent similar to the English agent.

• Add more languages to the agent's capabilities.

• Improve the interactive map and add support for transportation options with focus on public transportation.

• Add more services to help underprivileged communities.

Made with ❤️ by:

• Mohammed Syed

• Mohammed Ibrahim Ahmed