Enterprise AI Security: From Blueprint to Implementation

A 2-day intensive building enterprise AI defenses across the full stack — from RAG pipelines and prompt injection through agent security, MCP hardening, and operational controls

Course Format and Delivery

Delivery Method: LiveOnline
Schedule: 2 days

All sessions are delivered live by an expert instructor in a fully interactive online environment.

*20% off for group bookings when booking 3 or more attendees from the same organization on the same course dates in the same transaction.

About this course

97% of breached organizations had no proper AI controls in place. CVE-2026-25592 turned Semantic Kernel prompt injection into full host RCE. Researchers demonstrated that 1% poisoned vectors contaminate 99.85% of RAG query results. AI agents are the fastest-growing attack surface in enterprise software, yet most organizations deploy them with ad hoc security at best.

This 2-day intensive equips enterprise teams with the knowledge and hands-on skills to secure AI systems across the full stack — from RAG pipelines and AI agents to MCP tool integrations and production operations. Participants work through a battle-tested six-layer security blueprint (identity, data boundaries, prompt defenses, model governance, audit trails, operational controls) while building real defenses against the attacks that matter most: document poisoning, prompt injection, agent manipulation, tool abuse, and data exfiltration. Every concept is reinforced with hands-on labs where attendees first observe attacks succeeding on vulnerable systems, then implement and verify layered defenses.

By the end, participants have a concrete implementation roadmap and a Monday-morning checklist for immediate action.

What you will need

Basic familiarity with Python (reading and running scripts). General understanding of how LLMs and AI applications work. A laptop with a modern browser (Chrome recommended) and GitHub account. No prior AI security experience required.

For software engineers, security practitioners, DevOps/platform engineers, engineering managers, architects, and technical leaders responsible for building or governing AI systems. Content is primarily technical but structured so that leaders and decision-makers gain the strategic context needed to fund, prioritize, and oversee AI security initiatives. Day 1 covers AI security foundations and data protection: the OWASP Top 10 for LLM Applications (2025), the six-layer enterprise security blueprint, policy engine design, RAG attack and defense (document poisoning, defense in depth), PII tokenization, multi-tenant context isolation, and prompt injection defense patterns (direct, indirect, canary tokens, behavioral boundaries). Day 2 covers agent security, MCP, and enterprise operations: supervisor multi-agent patterns with budget enforcement, agent manipulation defenses, MCP authentication/authorization/per-tool scopes, MCP defense in depth (rate limiting, input validation, output sanitization), model governance and registry, structured audit trails, operational controls (kill switches, cost caps), and a capstone applying the full six-layer blueprint. 16 hands-on labs across both days, all running in GitHub Codespaces. Approximately 35% lecture, 65% hands-on labs.

Topics Covered

Day 1: AI Security Foundations & Data Protection

Time	Topic	Type
9:00–9:30	Welcome, threat landscape overview, environment setup	Setup
9:30–10:15	Why AI security is different, real-world incidents, the implementation gap	Lecture
10:15–11:00	OWASP Top 10 for LLMs, six-layer blueprint, policy engine framework	Lecture
11:00–11:15	Break	Break
11:15–11:45	Lab 1: AI Threat Modeling with OWASP Top 10	Hands-on
11:45–12:15	Lab 2: Policy Engine Decision Making	Hands-on
12:15–1:15	Lunch	Break
1:15–2:00	RAG security: attack surface, document poisoning, defense architecture	Lecture
2:00–2:30	Lab 3: RAG Attack — Exploiting a Vulnerable System	Hands-on
2:30–3:00	Lab 4: RAG Defense — Building Security Hardening	Hands-on
3:00–3:15	Break	Break
3:15–4:00	Context isolation, data boundaries, PII tokenization, multi-tenant security	Lecture
4:00–4:30	Lab 5: PII Detection & Tokenization Pipeline	Hands-on
4:30–5:00	Lab 6: Multi-Tenant Context Isolation	Hands-on
5:00–5:30	Prompt injection defenses: direct, indirect, four defense techniques	Lecture
5:30–6:00	Lab 7: Direct Prompt Injection — Attack & Defense	Hands-on
6:00–6:30	Lab 8: End-to-End Security Pipeline	Hands-on
6:30–6:45	Day 1 recap, Day 2 preview	Discussion

Day 2: Agent Security, MCP & Enterprise Operations

Time	Topic	Type
9:00–9:15	Day 1 review and questions	Discussion
9:15–10:00	Agent security: supervisor pattern, budget enforcement, threat model	Lecture
10:00–10:30	Lab 1: Supervisor Multi-Agent with Budget Enforcement	Hands-on
10:30–11:00	Lab 2: Securing Agents Against Manipulation	Hands-on
11:00–11:15	Break	Break
11:15–12:00	MCP security: authentication, authorization, tool isolation, defense in depth	Lecture
12:00–12:30	Lab 3: MCP Auth, Authorization & Per-Tool Scopes	Hands-on
12:30–1:00	Lab 4: MCP Defense in Depth — Rate Limiting, Validation & Sanitization	Hands-on
1:00–2:00	Lunch	Break
2:00–2:45	Model governance pipeline, registry, compliance; audit trails and observability	Lecture
2:45–3:15	Lab 5: Model Registry & Governance Gates	Hands-on
3:15–3:45	Lab 6: Structured Audit Trails for AI Systems	Hands-on
3:45–4:00	Break	Break
4:00–4:45	Operational controls: kill switches, cost caps, canary deployments, red teaming	Lecture
4:45–5:15	Lab 7: Operational Controls — Kill Switches & Cost Caps	Hands-on
5:15–6:15	Lab 8: Capstone — Enterprise AI Security Assessment	Hands-on
6:15–6:45	Implementation roadmap, Monday morning checklist, wrap-up	Discussion

Brent Laster

Facilitator

Brent Laster is a global trainer, author, speaker, and founder/president of Tech Skills Transformations LLC. He helps enterprise teams adopt modern software practices in AI engineering, AI-assisted development, DevOps, automation, and secure software delivery. He is the author of Learning GitHub Copilot, Learning GitHub Actions, Professional Git, and Jenkins 2: Up and Running, as well as multiple online and live training programs for companies such as O'Reilly. In addition to AI expertise, Brent brings more than 25 years of experience in software development, management and technical leadership, DevOps, release engineering, and open-source technologies. He regularly presents and conducts workshops at industry conferences and for private clients.