AI Governance Frameworks 🏛️

Welcome to the AI Governance Frameworks repository. This project represents a hands-on learning journey and an educational design exercise dedicated to exploring how abstract AI policies can be translated into structured engineering requirements.

Using high-stakes automated HR recruitment and resume screening as a foundational anchor (inspired by historical precedents like the Amazon hiring system case study), this repository demonstrates how to break down complex global regulations into logical, auditable blueprints.

---

📂 Repository Architecture

This repository splits the AI risk lifecycle into two complementary, conceptual artifacts:

          ┌────────────────────────────────────────┐
          │ 1. AI Risk Checklist (Hiring System)   │
          │    - Diagnostic Audit Tool             │
          │    - Evaluates & Scores System Risks   │
          └───────────────────┬────────────────────┘
                              │
                              ▼
          ┌────────────────────────────────────────┐
          │ 2. Resume Screening Governance Matrix │
          │    - Operational Lifecycle Blueprint   │
          │    - Continuous Control Loop (NIST)    │
          └────────────────────────────────────────┘

1. AI Risk Checklist — Hiring System

• Purpose: A diagnostic audit tool designed to turn qualitative risk evaluation into a structured, quantitative checklist.
• Core Logic: Uses an objective scoring system (Yes = 0, Not Sure = 1, No = 2) across 4 critical risk dimensions to sort an AI application into distinct risk tiers (Low, Medium, High).
• Case Study Validation: Pre-populated using a retroactive audit of the defunct historical Amazon hiring tool, resulting in a maximum severity score of 16/16 (High Risk) to demonstrate how structured checks flag systemic flaws before deployment.

2. Resume Screening Governance Framework

• Purpose: An end-to-end operational blueprint outlining a continuous organizational control loop.
• Core Logic: Maps the four core functional layers of the NIST AI Risk Management Framework (AI RMF 1.0) specifically to an internal Machine Learning resume screening pipeline.
• Regulatory Mapping: Connects technical engineering actions (like bias tracking, drift measurement, and logging) directly to mandatory compliance clauses under the EU AI Act (High-Risk AI Systems under Annex III).

---

🛡️ Core Pillars & Regulatory Alignment

Both documents explore how to operationalize high-level trustworthy AI principles by explicitly mapping them to regional statutes:

• Data Equity & Representation: Auditing training datasets for historical bias, proxy features, and demographic equity (EU AI Act Art. 10).
• Technical Stability & Monitoring: Designing frameworks to track performance degradation and feature/data drift over time (EU AI Act Art. 15).
• System Traceability: Ensuring robust, automated logging mechanisms are built into the pipeline infrastructure (EU AI Act Art. 12).
• Human Oversight & Governance: Defining clear executive accountability (CHRO/CRO ownership) and protocols to prevent human confirmation bias (EU AI Act Art. 14).

---

🛠️ How to Utilize This Repository

This repository is built as a study guide and a practical conceptual reference, rather than a production software application.

• For Learners & Peer Auditors: Use the AI Risk Checklist to see how a historical failure can be broken down systematically into discrete, quantifiable evaluation categories.
• For Aspiring Applied Scientists & Engineers: Treat the Governance Framework Matrix as a structural exercise to understand the organizational, data, and compliance layers required to support a machine learning pipeline.
• For Anyone Navigating AI Regs: Use this repository as a practical translation exercise to visualize how abstract regulatory text maps to a tangible, real-world domain.

---

Developed as an educational blueprint to transition theoretical trustworthy AI frameworks into structured, auditable concepts.