An integrated operational intelligence prototype combining demand forecasting, customer RFM segmentation, seller anomaly detection, and document-grounded policy retrieval.
E-commerce merchants frequently manage fragmented operational data: internal return policies and shipping guidelines live in disconnected documents, while demand forecasts and anomaly detection scripts remain isolated in offline notebooks without standard APIs for operations teams.
Engineered a unified demonstration prototype that integrates Prophet time-series demand forecasting, historical RFM customer segmentation, Isolation Forest seller anomaly detection, and a context-bounded ChromaDB vector search RAG engine behind an asynchronous FastAPI REST layer and interactive Streamlit UI.
| Component Layer | Role & Implementation |
|---|---|
| FastAPI Microservice Layer | Asynchronous REST endpoints (/health, /predict, /api/v1/query) with strict Pydantic input/output schemas |
| Predictive ML Engine | Prophet (90-day demand forecast), Gradient Boosting (churn classification), and Isolation Forest (seller anomalies) |
| RAG Intelligence Core | ChromaDB vector store with semantic retrieval and grounded context injection for internal policies |
| Streamlit Dashboard | 6-page operational UI (Overview KPIs, Demand Forecast, Customer Segments, Anomalies, AI Reports, AI Assistant) |
| Containerization | Docker Compose orchestration ensuring consistent microservice dependency management |
Get in Touch
Tell me about your agency challenge, spreadsheet bottlenecks, or workflow needs. I will review the details and suggest a practical approach.
Prefer direct communication over the form? Send an email or connect via WhatsApp: