How vikash-vk/ai-ecommerce-operations-platform works
ShopKart Operations Assistant is a unified Streamlit interface for business-data questions, uploaded-PDF search, membership prediction, and BI report access. The main flow is a user request through the app’s agent to a selected tool, then to SQL generation/execution, document retrieval, or model inference. A separate data-preparation stage cleans CSVs and builds the SQLite database. The dashboard is provided as Power BI files. Relationships reflect the excerpts and README; unsampled dashboard internals and model training details are not inferred.
Main entry points: app.py Streamlit assistant · database_query.py Database query tool · connection.py SQLite connection · create_database.py Database builder
Components 19
Assistant interface
- Streamlit assistant
app.py - LLM configuration
config.py
Database analytics
- Database query tool
tools/database_query.py - Schema discovery
database/schema.py - SQL generation
database/llm.py - SQL execution
database/execute.py - SQLite connection
database/connection.py - ShopKart database
database/shopkart.db
Document search
- PDF search tool
tools/pdf_search.py - PDF indexing
rag/embeddings.py - PDF documents
data/pdfs
Membership prediction
- Membership prediction
tools/membership_prediction.py - Trained model
models/membership_model.pkl
Data and reporting
- Database builder
database/create_database.py - CSV cleaning
database/etl.py - Raw business CSVs
data/raw - Database schema
database/schema.sql - Power BI reports
Dashboard
Other components
- Business user
Connections 21
- Business user → Streamlit assistant: asks
README.md - Streamlit assistant → LLM configuration: configures model
app.py - Streamlit assistant → Database query tool: dispatches queries
app.py - Streamlit assistant → PDF search tool: dispatches search
app.py - Streamlit assistant → Membership prediction: dispatches prediction
app.py - Streamlit assistant → PDF indexing: indexes uploads
app.py - Database query tool → Schema discovery: gets schema
tools/database_query.py - Database query tool → SQL generation: generates SQL
tools/database_query.py - Database query tool → SQL execution: executes SQL
tools/database_query.py - Schema discovery → SQLite connection: opens connection
database/schema.py - SQL execution → SQLite connection: opens connection
database/execute.py - SQL generation → LLM configuration: uses model settings
database/llm.py - SQLite connection → ShopKart database: connects to
database/connection.py - PDF search tool → PDF indexing: searches vectors
tools/pdf_search.py - PDF indexing → PDF documents: indexes uploaded files
- Membership prediction → Trained model: loads model
tools/membership_prediction.py - Database builder → CSV cleaning: loads cleaned data
database/create_database.py - Database builder → SQLite connection: creates database
database/create_database.py - Database builder → Database schema: executes schema
database/create_database.py - CSV cleaning → Raw business CSVs: reads CSVs
database/etl.py - Streamlit assistant → Power BI reports: offers report downloads
app.py