banking-risk

Production-grade synthetic retail banking data generator that simulates customers, accounts, transactions, loans, cards, churn events, and customer lifecycles. Generates realistic, relational datasets at scale for Data Engineering, Analytics, Machine Learning, and MLOps projects.

Early-warning models that correctly predicted the 2008 and 2020 banking crises — signal-based NSR + Probit regression on 33 macroeconomic indicators, 74.5% out-of-sample accuracy. IIT Kanpur research internship.

Synthetic banking AI risk management control tower demonstrating AI inventory, risk assessment, mitigation planning, controls, evidence tracking, ownership, and framework mapping.

public-data macro-financial stress test for JPMorgan Chase, integrating credit risk, regulatory capital, liquidity analysis, DFAST-style scenarios and reproducibility documentation.

A full-stack Banking Management System built with Flask and MySQL, integrated with Machine Learning for real-time transaction fraud detection, risk scoring, complete CRUD management (Customers, Accounts, Loans, Cards, Employees), interactive analytics, and automated PDF/Excel reports.

Interest Sensitivity (IS) Gap–based IRRBB model to analyze Net Interest Income (NII) impact under upward and downward interest rate shocks, implemented in Python with FRM-aligned methodology.

Public FDIC banking-risk data platform with SQL modelling, data-quality controls, peer benchmarking, chronological validation, and an explainable Streamlit dashboard.

Streamlit MVP for SME loan pre-screening with credit scoring, KYC checks, and explainable Go/Review/Decline decisions.