🛡️ Open Source Risk Framework

A modular, production-ready, open-source risk management framework for detecting and analyzing Financial Crime, Operational Risk, Market & Credit Risk, Liquidity Stress, and AI governance with built-in Model Context Protocol (MCP) support.

---

🎯 Project Purpose

This framework empowers compliance teams, risk analysts, data scientists, and developers to:

• Identify and mitigate fraud, AML, and KYC risks
• Quantify and explain credit, market, and liquidity risk
• Automate risk modeling with explainable AI
• Ensure model transparency with Model Context Protocol (MCP)
• Customize & extend modules with minimal overhead

---

🏗️ Architecture Overview

risk_framework/
├── ingestion/         # Loaders for CSV, AWS, GCP, Azure
├── models/            # Domain-specific ML models (modular)
│   ├── credit_risk/
│   ├── market_risk/
│   ├── operational_risk/
│   ├── liquidity_risk/
│   ├── fincrime_aml_kyc/
│   ├── regulatory/
│   └── utils/
├── evaluation/        # SHAP explainability, bias audit
├── reporting/         # Report generation utilities
├── mcp/               # Model Context Protocol (YAML templates + validators)
dashboard/             # Streamlit UI
examples/              # Jupyter notebooks
tests/                 # Unit tests

---

🔍 Key Features

• ✅ Modular ML Models for various risk domains
• 📜 MCP (Model Context Protocol) templates and validation
• 🔍 Anomaly detection, graph analysis, NLP, and SHAP
• 📊 Streamlit Dashboard to visualize flagged transactions
• 🧪 Notebook Examples for usage in real-world scenarios
• 📦 Easily extensible with plug-and-play components

---

📦 Installation

git clone https://github.com/Souptik96/open-source_risk_framework.git
cd open-source_risk_framework
pip install -r requirements.txt

---

🚀 Quick Start (Streamlit)

streamlit run open-source-risk-framework/dashboard/app.py

Upload a transaction .csv file and let the Isolation Forest model flag potential anomalies!

---

🧠 Example Usage (Notebook)

See: examples/financial_crime_example.ipynb

from risk_framework.models.fincrime_aml_kyc.isolation_forest import IsolationForestModel
model = IsolationForestModel()
model.fit(data)
predictions = model.predict(data)

---

📁 Model Context Protocol (MCP)

Each model supports machine-readable metadata via YAML to ensure auditability, reproducibility, and trust.

model_name: IsolationForestModel
author: Souptik Chakraborty
version: 1.0
data_sources:
  - source: transactions.csv
  - schema: { amount: float, merchant_id: str, country: str }
...

---

🧪 Testing

pytest tests/

---

🤝 Contributing

We welcome community contributions! Please see CONTRIBUTING.md for guidelines.

---

📜 License

Apache 2.0. See LICENSE.

---

🌐 Stay Connected

• LinkedIn
• Project Repo
• Contact: souptikchakraborty@gmail.com

---

⭐ Star the repo if you like the project or plan to use it!