Course – AI Security, Governance & Operations
AI Security & Governance: A Practitioner’s Program
3 Days (18–24 Hrs)Intermediate–AdvancedClassroom / Live VirtualHands-On Labs & Capstone
Duration
3 Days / 18-24 Hrs
Level
Intermediate-Advanced
Format
Classroom / Live Virtual
Focus
Governance · Security · Operations
Program Overview
- This program equips your teams with the knowledge and hands-on skills to secure, govern, and scale AI across the enterprise — covering emerging threats, regulatory requirements, secure development practices, and real-world implementation, backed by practical labs and case studies.
- Across three progressive days, you’ll move from core AI concepts and governance fundamentals, through secure development and threat modeling, into advanced architectures like RAG and autonomous agents — closing with a hands-on capstone that ties it all together.
Who Should Attend
What You’ll Walk Away With
- A common language for AI, ML, LLMs, and agentic systems across technical and non-technical teams
- Confidence navigating the regulatory landscape — EU AI Act, NIST AI RMF, ISO/IEC 42001, OWASP, MITRE ATLAS
- A repeatable method for threat-modeling AI systems using MAESTRO and STRIDE
- Hands-on skill in secure AI development, guardrails, and cost/token optimization
- A clear view of RAG and agent architectures — and how to keep the data behind them safe
- Operational readiness to monitor, observe, and respond to AI incidents in production
- A capstone experience applying everything to a real-world scenario
Curriculum
Day 1AI Foundations, Standards & Governance
Module 1
Definitions & Terminology
- Core AI/ML vocabulary — machine learning, deep learning, LLMs, tokens, embeddings, inference, fine-tuning, hallucinations, and prompt engineering.
Module 2
Types of AI & AI Applications
- Overview of generative AI, predictive AI, classification, NLP, computer vision, and real-world enterprise use cases.
Module 3
AI Agents
- What AI agents are, how they differ from simple LLM calls — autonomy, tool use, planning loops, memory, and multi-agent architectures.
Module 4
Standards & Regulatory Landscape
- EU AI Act, NIST AI RMF, ISO/IEC 42001, Executive Orders, and sector-specific requirements.
Module 5
Frameworks (OWASP, ISO, RMF, MITRE ATLAS)
- OWASP Top 10 for LLMs, ISO/IEC 42001, NIST AI RMF, and MITRE ATLAS.
Module 6
Explainability (XAI)
- Why AI decisions need to be interpretable — SHAP, LIME, attention visualization, and regulatory requirements.
Module 7
Auditability & Accountability
- Logging AI decisions, audit trails, model versioning, and lineage tracking.
Module 8
AI Data Readiness (AIDR)
- Data quality, governance, bias detection, and labeling standards.
Module 9
Responsible AI Principles
- Fairness, transparency, privacy, safety, inclusiveness, and reliability.
Day 2AI Development, Consumption & Threat Modeling
Module 1
AI Development Lifecycle (AIDLC)
- Problem framing, data preparation, model selection, training, evaluation, deployment, and decommissioning.
Module 2
AI as a Consumer (Copilot, ChatGPT, etc.)
- Prompt hygiene, data leakage prevention, acceptable use policies, and output validation.
Module 3
Token Optimization
- Context window management, prompt compression, chunking, caching, and model/tier selection to reduce latency and cost.
Module 4
AI in Applications (Integration Patterns)
- APIs, SDKs, orchestration layers, and architecture patterns (retrieval-augmented, agentic, fine-tuned).
Module 5
AI Threat Modeling
- Prompt injection, data poisoning, model theft, and supply chain risks — mapped to STRIDE/MITRE ATLAS.
Module 6
MAESTRO Framework
- OWASP’s AI security threat modeling methodology and its layers.
Module 7
Secure AI Development Practices
- Input/output validation, guardrails, sandboxing, and secrets management.
Day 3Advanced — RAG, Agents, Secure Data & Operations
Module 1
Retrieval-Augmented Generation (RAG)
- Chunking strategies, embedding models, vector databases, retrieval pipelines, and re-ranking.
Module 2
AI Agents (Advanced)
- Tool calling, chain-of-thought reasoning, guardrails, human-in-the-loop, and multi-agent orchestration.
Module 3
Secure Data Modeling for AI
- Access controls on vector stores, data classification, PII/PHI filtering, and differential privacy.
Module 4
AI Operations (AIOps / MLOps)
- Model deployment pipelines, A/B testing, canary rollouts, model registry, and versioning.
Module 5
Monitoring & Observability
- Drift detection, performance degradation, hallucination tracking, and cost monitoring.
Module 6
Incident Response for AI Systems
- Containment strategies, bias incidents, data breach through AI, and communication playbooks.
Module 7
Capstone Exercise
- Threat model a RAG-based AI application, identify risks, apply controls from Days 1-3, and present findings.
Delivery Details
- Delivered as classroom or live virtual instructor-led — 3 full-day sessions, or split across 6 half-days to fit your team’s schedule
- Includes real-world case studies and a hands-on capstone applying governance, security, and operations controls to a RAG-based scenario
- Labs and exercises are illustrative and may vary by trainer approach and participant technical background