San Francisco Small Business Intelligence Platform

A comprehensive hackathon project providing AI-powered intelligence for small businesses in San Francisco. The platform includes three main features: Risk Prediction Engine, Lease Negotiation Intelligence, and City Fee Analysis. This project aims to empower local entrepreneurs with data-driven insights.

🎯 Features

1. Risk Prediction Engine

• ML-powered risk assessment using Random Forest and Gradient Boosting models
• Predicts business failure risk based on:
  • Business age and activity
  • Permit history
  • Code enforcement complaints
  • Location status
• Real-time risk alerts and recommendations

2. Lease Negotiation Intelligence

• Market analysis by neighborhood
• AI-generated negotiation strategies using Nemotron LLM
• Comparable lease rate analysis
• Counter-proposal generation
• Customized talking points and concession suggestions

3. City Negotiation Intelligence

• Fee analysis and breakdown for permits and licenses
• Waiver opportunity identification (small business, nonprofit, new business)
• Compliance requirement checklist with AI-powered advisor
• Renewal schedule tracking

🛠️ Tech Stack

• Python 3.10+
• DGX Spark with Nvidia Nemotron LLM (via OpenAI-compatible API)
• SF.gov Open Data API for business, permit, and complaint data
• Streamlit for interactive web UI
• scikit-learn, pandas, numpy for ML and data processing
• OpenAI client library (configured for local Nemotron endpoint)

📁 Project Structure

sf-business-intelligence/
├── data/
│   ├── raw/              # Raw data from SF.gov API
│   ├── processed/         # Cleaned and merged datasets
│   └── models/            # Trained ML models
├── src/
│   ├── data_pipeline/     # Data download, cleaning, merging
│   ├── risk_engine/       # Risk prediction model and alerts
│   ├── lease_intelligence/ # Lease market analysis and AI strategies
│   ├── city_intelligence/  # Fee analysis and compliance
│   └── utils/             # Config and Nemotron client
├── app/
│   ├── streamlit_app.py   # Main Streamlit app
│   └── pages/             # Feature pages
├── notebooks/             # Data exploration notebooks
├── requirements.txt
├── setup.py
└── README.md

🚀 Getting Started

Prerequisites

• Python 3.10 or higher
• Access to SF.gov Open Data API (optional app token)
• Nemotron LLM running locally or remotely (for AI features)

Installation

1. Clone or navigate to the project directory:

   cd sf-business-intelligence

2. Create a virtual environment:

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

3. Install dependencies:

   pip install -r requirements.txt

4. Set up environment variables (optional): Create a .env file in the project root:

   SF_DATA_APP_TOKEN=your_sf_data_token
   NEMOTRON_BASE_URL=http://localhost:8000/v1
   NEMOTRON_API_KEY=local-nemotron-key
   NEMOTRON_MODEL=nvidia/nemotron-4-340b-instruct

Running the Application

1. Start the Streamlit app:

   streamlit run app/streamlit_app.py

2. Access the app: Open your browser to http://localhost:8501

3. Navigate features:

   • Risk Dashboard: Train models and analyze business risk
   • Lease Negotiation: Get market analysis and AI strategies
   • City Fees: Analyze fees and check compliance

📊 Data Sources

The platform uses SF.gov Open Data API:

• Business Registry (rqzj-sfat): Business registration and license data
• Building Permits (p4e4-5k3y): Permit applications and approvals
• Code Enforcement Complaints (ktji-gkfc): Complaint records

🤖 AI Features (Nemotron LLM)

The platform uses Nemotron LLM for:

• Generating lease negotiation strategies
• Creating counter-proposals
• Providing compliance requirement advice

Note: Ensure Nemotron is running and accessible at the configured endpoint. The OpenAI client library is used with a local Nemotron endpoint.

🔧 Configuration

Edit src/utils/config.py to customize:

• SF.gov API endpoints and datasets
• Nemotron LLM endpoint and model
• Risk thresholds
• Data directories

📝 Usage Examples

Risk Prediction

1. Navigate to "Risk Dashboard"
2. Train or load a risk model
3. Enter business information or analyze trends
4. Review risk scores and alerts

Lease Negotiation

1. Navigate to "Lease Negotiation"
2. Enter business and lease information
3. Get market analysis
4. Generate AI-powered negotiation strategy
5. Create counter-proposals

City Fees

1. Navigate to "City Fees"
2. Enter business information
3. Select required permits
4. Review fee breakdown and waiver opportunities
5. Check compliance requirements

🧪 Development

Running Tests

pytest tests/

Data Exploration

jupyter notebook notebooks/exploration.ipynb

Code Formatting

black src/
flake8 src/

📄 License

This project is created for hackathon purposes.

🙏 Acknowledgments

• SF.gov Open Data for providing public datasets
• Nvidia for Nemotron LLM
• Streamlit for the UI framework

🔮 Future Enhancements

• Real-time data updates
• Integration with more SF.gov datasets
• Advanced ML models (deep learning)
• Mobile app version
• API endpoints for programmatic access
• Multi-city support

---

Built for SF Small Businesses | Powered by DGX Spark & Nemotron LLM