Retail Banking EDA & Transaction Analytics Platform | Capstone Project

Retail Banking EDA & Transaction Analytics Platform is an intermediate-level capstone that takes you through the full data lifecycle of a retail banking system. You start with raw nested JSON transaction files and end with live analytics dashboards.

First, you build an ingestion layer that uses SHA-256 hashing so the same file is never loaded twice. Next, you run an automated 8-step EDA pipeline with Pandas and NumPy to flatten, clean and analyze the data. The cleaned data goes into a 14-table PostgreSQL database designed as a star schema. On top of the database, you add fraud detection with a 0–100 risk score, geo-anomaly and VPN flags, and tracking for 10+ Indian regulatory compliance flags, including AML, KYC, FATCA, PEP, CTR, STR and RBI reporting. Finally, a FastAPI backend serves the results to 7 interactive dashboards built with Chart.js.

Banking is one of the highest-paying domains for data roles. By the end, you’ll have a portfolio project that shows real BFSI domain knowledge, careful handling of data integrity, and the engineering practices FinTech employers look for.

By the end of this capstone, you will be able to:

• Ingest nested JSON banking data, using SHA-256 hashing to detect and skip duplicate files.
• Build an automated 8-step EDA pipeline with Pandas and NumPy: flattening, type casting, handling missing values, cleaning and analysis.
• Explore financial data from multiple angles, including spending patterns, channel usage, customer segments and trends over time.
• Design a 14-table PostgreSQL database as a star schema, with fact and dimension tables, managed through SQLAlchemy and Alembic.
• Detect fraud and score transaction risk on a 0–100 scale, using rules for geo-anomalies, VPN use, high-value transactions and unusual behaviour.
• Track regulatory compliance flags: AML, KYC, FATCA, PEP, CTR, STR and RBI reporting requirements.
• Validate banking identifiers such as PAN, Aadhaar, SWIFT and IBAN, and apply basic data masking for sensitive fields.
• Build REST APIs with FastAPI to serve transaction, risk and compliance analytics, with filtering and pagination.
• Build 7 interactive dashboards with Vanilla JavaScript and Chart.js, covering transaction volumes, fraud trends, risk distribution and compliance status.
• Connect the whole pipeline end to end, from raw data to EDA to the database to dashboards, as a production-style data engineering project.