cAIge Training Program
Certified AI Guardrail Engineer (cAIge)
Version: 1.0 Maintained by: caige.org
Program Overview
AI guardrails are the technical controls that prevent AI systems from producing harmful content, being exploited by adversarial users, leaking sensitive data, or taking unauthorized actions. A guardrail engineer designs, builds, and validates these controls — specifically for applications powered by large language models (LLMs).
This training program prepares candidates for the Certified AI Guardrail Engineer (cAIge) examination. It covers all five domains of the cAIge competency matrix with text-based instruction, real-world examples, code samples, and architectural diagrams aligned to the exam blueprint.
The program is vendor-agnostic. It teaches concepts, patterns, and engineering thinking — not product-specific configuration. You will learn to design, implement, test, and maintain AI guardrails across any technology stack.
Prerequisites
Before starting the training modules, you should have:
- 1-2 years working with AI/ML systems in any capacity
- Basic understanding of software architecture and API design
- Familiarity with at least one programming language (Python examples are used throughout)
- Exposure to security concepts (you don’t need to be a security specialist)
Module 1 (AI Foundations) covers the foundational AI knowledge required for the certification — including neural networks, the transformer architecture, LLM inference, training pipelines, and production AI system architecture. If you are already comfortable with these concepts, you can move quickly through Module 1 and focus your time on Modules 2-5.
Training Modules
| Module | Domain | Weight | Estimated Time |
|---|---|---|---|
| Module 1: AI Foundations | Domain 1 | 15% | 5-6 hours |
| Module 2: Understanding Failure Modes | Domain 2 | 15% | 3-4 hours |
| Module 3: Architecting Guardrails | Domain 3 | 25% | 6-8 hours |
| Module 4: Implementing Guardrails | Domain 4 | 25% | 6-8 hours |
| Module 5: Validating Guardrails | Domain 5 | 20% | 5-6 hours |
Total estimated study time: 25-32 hours
How to Use This Training
- Study in order. The five modules follow the guardrail engineering lifecycle: understand the technology → understand the failures → design defenses → build them → validate them. Each module builds on the previous ones.
- Read actively. Don’t skim. The exam tests applied understanding, not memorization. When you encounter a concept, think about how you’d apply it in your own work.
- Run the code examples. Modules 3-5 include Python code patterns for detection pipelines, testing frameworks, and monitoring systems. Running these examples will deepen your understanding.
- Build something. The best preparation combines this training with hands-on experience. Set up a test environment and implement the guardrails discussed in each module.
- Review the competency matrix. After completing all modules, review the competency matrix and honestly assess whether you can demonstrate each listed skill. Focus additional study on gaps.
Supporting Materials
- Competency Matrix — Complete breakdown of knowledge areas and skills per domain
- Exam Blueprint — Exam structure, question types, and detailed objectives
Exam Information
| Parameter | Detail |
|---|---|
| Format | Multiple choice, multiple select, and scenario-based |
| Questions | 75 scored + 5 unscored pilot questions |
| Time | 120 minutes |
| Passing score | 70% (53 of 75) |
| Prerequisites | None required to sit for the exam; training program completion strongly recommended |
| Delivery | Online proctored |
Version History
| Version | Date | Changes |
|---|---|---|
| 1.0 | 2026-03-07 | Initial release |