Building and Securing RAG and Agentic AI Systems
Retrieval augmented generation (RAG) and agentic systems are where most enterprise AI is heading, and where most of the new risk lives. This course in Toronto has you build them hands on, from basic through contextual and agentic, and then secure what you built.
You will see how a poisoned document becomes indirect prompt injection, why a shared vector store leaks between tenants, and how giving an agent tools quietly hands an attacker the same tools. Then you constrain all three.
- Duration
- 8 weeks
- Format
- Live online, instructor-led
- Prerequisites
- AI Security Principles and Practices: GenAI and LLM Defence
- Class size
- Capped at 14 learners
What you will be able to do
- Build a retrieval augmented generation pipeline end to end
- Explain how retrieved content becomes indirect prompt injection
- Isolate a vector store so one tenant cannot read another's data
- Constrain an agent's tools to least privilege and safe actions
- Add grounding and output checks that reduce hallucination risk
- Decide which agent actions require a human in the loop
Course outline
6 modules
- Embeddings, vector stores, and retrieval
- Chunking and why it affects both quality and security
- Basic, contextual, and agentic RAG compared
- Where untrusted content enters the pipeline
What you need before you start
- You have completed AI Security Principles and Practices or have equivalent knowledge
- You can program in Python at an intermediate level
- You understand APIs and basic web application concepts
- You have used a large language model API before, even briefly
Who this course is for
- Engineers building RAG or agentic AI features
- Security engineers reviewing AI systems for production
- ML and platform teams adding retrieval to their products
- Technical leads deciding how much autonomy to give an agent
Questions about this course
More in ai security
CoreAI Security Principles and Practices: GenAI and LLM Defence
The foundational AI security course: threat model generative AI, defend against prompt injection, and set pragmatic policy for an enterprise rolling out AI.
8 weeks · live online
Contact for pricingView outline
CoreAdversarial AI: Penetration Testing AI Systems
Penetration test AI systems themselves: model attacks, jailbreak methodology, data extraction, and reporting findings that are probabilistic, not binary.
6 weeks · live online
Contact for pricingView outline
CoreAI Powered Security Automation with Python
Automate security operations with Python, large language models, and the Model Context Protocol, building agentic tools that fail safely and keep a human in the loop.
8 weeks · live online
Contact for pricingView outline
Ready to launch your cyber security career?
Join the next live online cohort. No experience required, just bring the curiosity.