EXPLORATORY DATA ANALYSIS

Financial Transaction Dataset

The Story

This project was born from the need to understand complex financial transaction patterns while providing meaningful insights through advanced statistical analysis and visualization techniques.

Key Features

• Data extraction from PostgreSQL databases
• Data cleaning and preprocessing
• Advanced outlier detection and treatment
• Feature engineering for users, cards, and transactions
• Interactive data visualizations with Matplotlib & Seaborn
• Export to SQL for downstream analytics and Power BI dashboards

Technologies

BRAIN TUMOR AI DETECTOR

Deep Learning for Medical Imaging

The Story

This project began as an exploration into applying deep learning to medical imaging. I wanted to solve a real-world problem: detecting brain tumors from MRI scans. The journey included data preparation, model development, evaluation using accuracy and precision metrics, and deployment into a FastAPI web application with Docker containerization.

Key Features

• Classification: Predicts tumor presence using InceptionV3 transfer learning
• Segmentation: U-Net model generates tumor mask overlays
• Web App: User-friendly FastAPI interface for upload and results
• Deployment: Fully containerized with Docker for cloud deployment
• Stateless API: Processes results without storing patient data
• Visualization: Returns overlay images with tumor highlighted in red
• Logging: Records predictions with timestamps for monitoring

Technologies