Enterprise E-Commerce Intelligence & Predictive Analytics (DEPI Graduation Project)

Developed an end-to-end modern data analytics platform transforming 13,000+ raw Amazon transactions (UK, DE, FR, IT, ES) into a centralized Single Source of Truth. This system bridges granular operations with executive strategy, providing real-time decision support for product intelligence and risk mitigation.

Data Engineering & Architecture

Engineered an automated pipeline in Microsoft Fabric using a Medallion Lakehouse architecture (Bronze, Silver, Gold). Leveraged Python (Pandas) for data remediation—handling missing values, deduplication, and metric standardization—before building an optimized Star Schema (3 Dimensions, 1 Fact) for Power BI.

AI & Predictive Analytics

Deployed an unsupervised Isolation Forest machine learning model to proactively detect abnormal sales patterns, revenue fluctuations, and operational anomalies (e.g., refund spikes). This AI integration isolates high-risk events and assigns anomaly scores, drastically reducing manual data inspection.

Business Intelligence & Dashboarding

Designed the UI/UX in Figma using strategic color psychology, then built an interactive, 7-page executive suite in Power BI. Core modules include:

Executive & SKU Tracking: 80/20 Profit Pareto analysis isolating top revenue drivers from margin-eroding "Bleeding SKUs."

Marketplace & Marketing: Regional mapping and Halo Effect quantification to measure PPC impact on organic sales.

Refunds & Promotions: Tracking revenue lost to returns and evaluating the true ROI of promo campaigns.

Anomaly Detection: Dedicated AI interface surfacing unusual product behaviors and daily trends.

Technical Stack

Business Intelligence: Power BI (DAX, Slicers, Interactive Navigation)

Machine Learning: Python (Isolation Forest for unsupervised anomaly detection)

UI/UX Design: Figma (Layouts, Color Palettes, Wireframing)