Real-Time Retail Analytics Platform built on the Medallion Architecture.
RetailSink is an end-to-end data platform that simulates retail operations (E-commerce, POS, Warehouse) and processes the resulting data through a multi-layered Lakehouse architecture to provide real-time business insights.
This project includes several detailed markdown files covering different aspects of the architecture and implementation:
| File | Purpose |
|---|---|
| documentation.md | Main Technical Docs: Detailed overview of the pipeline scripts, design choices (ELT, DuckDB, Delta Lake), and SCD Type 2 implementation. |
| data_flow.md | Architecture Visualization: Mermaid diagrams and descriptions of data movement from Landing to Gold layers. |
| notes.md | Developer Notes: Data schema mappings, simulator fault probabilities, and internal phase-wise planning. |
| PowerBI_Demo_Video.md | Product Demo: External drive link to a video demonstrating the PowerBI dashboard implementation. |
To run the entire RetailSink platform locally:
Run the master orchestrator to start data generation and the backend server:
python main.pyIn a new terminal, navigate to the UI directory and start the Vite dev server:
cd ui
npm run devThe dashboard will be available at the URL displayed in your terminal.
The analytical backend runs in the background. You can access the interactive API docs at the /docs endpoint of the API server.