🍟 QSR Dashboard — AI-Powered Financial Insights for Restaurants

An interactive, AI-enhanced dashboard for visualizing and analyzing Quick Service Restaurant (QSR) financial performance across stores, regions, and time. Built with Dash and powered by a custom Retrieval-Augmented Generation (RAG) system for natural language querying.

---

Click to view Mermaid code

flowchart LR
    A["User Query"] --> B["Retrieve Chunks from FAISS"]
    B --> C["Chunked Financial Data - JSON"]
    C --> B
    B --> D["LLM via Together AI"]
    D --> E["Natural Language Answer"]

    subgraph Embedding Store
        C
        B
    end

    style A fill:#f0f8ff,stroke:#333,stroke-width:1px
    style E fill:#e6ffe6,stroke:#333,stroke-width:1px
    style D fill:#fff0f5,stroke:#333,stroke-width:1px

Loading

---

🚀 Key Features

• Dynamic Financial Dashboard
  Visualize KPIs like revenue, cost of sales, and EBITDA across periods, regions, area coaches, and stores.

• Modular Views

  • Summary view with top KPIs and waterfall chart
  • Revenue, cost of sales, and payroll breakdowns
  • Store-level and regional comparisons

• Ask AI: RAG-Based Chatbot

  • Natural language Q&A over store financial data
  • Powered by Together AI and FAISS for accurate, context-aware answers
  • Built-in chunking and retrieval on structured JSON data

• Fully Interactive

  • Dropdown filtering
  • Responsive charts and tables
  • Auto-generated insights based on selections

---

🧠 Tech Stack

• Frontend & Framework: Dash, Plotly, Dash Bootstrap Components
• Backend: Flask, Gunicorn, FAISS
• LLM Integration: Together AI via API
• Embedding: Sentence Transformers
• Data Handling: Pandas, NumPy
• Deployment: Procfile + Gunicorn
• Env Config: python-dotenv

---

📦 Project Structure

📁 Data/
│ ├── all_store_data_region.json
│ └── rag_chunks/
│ ├── store_chunks.jsonl
│ ├── financial_index.faiss
│ ├── texts.pkl
│ └── metadata.pkl
📁 callbacks/
📁 components/
📁 rag_together/
📁 utils/
├── main.py
├── requirements.txt
├── Procfile
├── .env
└── README.md

---

⚙️ Setup & Installation

1. Clone the Repository

git clone https://github.com/aryanj10/dashboard.git
cd dashboard

2. Create and Activate Virtual Environment

python -m venv venv
source venv/bin/activate    
# On Windows: venv\Scripts\activate

3. Install Requirements

pip install -r requirements.txt

4. Add Environment Variable

Create a .env file:

TOGETHER_API_KEY=your_together_api_key

5. Preprocess Data for RAG

python rag_together/rag_code.py

If needed, create the FAISS index:

python rag_together/Faiss_indexing.py

6. Run the App

➤ Development

python main.py

Visit: http://localhost:7000/

➤ Production (Gunicorn)

gunicorn main:server

---

AI Chat — How It Works

• Financial JSON is converted to text chunks.

• Chunks are embedded using Sentence Transformers.

• FAISS indexes embeddings for fast retrieval.

• On a query, the most relevant chunks are retrieved.

• Together AI LLM answers the question based on those chunks.

---

🙌 Acknowledgments

• Together AI for free LLM API access

• Dash community for open-source components

• FAISS and Sentence Transformers for powering semantic search

---