Overview
The AI security, safety and governance interview, answered. Eleven volumes start with threat modeling for AI systems and move through where AI systems break, how attacks actually happen, defenses and guardrails, and secure system design for production. They then cover governance, ethics and compliance, real-world use cases, agent and protocol security, privacy engineering for LLM systems, red teaming and guardrail engineering, and incident case studies. Hands-on questions hand you the prompt, config or log to analyze. Every answer is written in the first person, and many carry a diagram.
What you will learn
- Threat-model LLM, RAG and agent systems end to end
- Explain prompt injection, data exfiltration and jailbreak techniques
- Design layered defenses and guardrails for production AI
- Secure agents, tools and protocols such as MCP
- Engineer privacy into LLM systems and their data
- Plan and run red teaming, and tune guardrails from the results
- Answer governance, compliance and responsible AI questions with specifics
- Analyze real incidents and say what would have prevented them
Included with the kit
- 113 written parts, yours for good
- 20h 4m of reading, measured not estimated
- Written for the advanced level
- Every future revision included
- UPI, cards and netbanking
Prerequisites
- Experience building or operating LLM applications
- Basic security concepts; AI-specific attacks are explained as they come up
Curriculum
11 sections · 113 parts · 20h 4m01Volume 0: Foundations - Threat Modeling for AI Systems5 parts · 85 min
Build a strong mental model of how AI systems work and where risks originate
- Foundations24 min
- Threat Modeling in Practice15 min
- Trust Boundaries for Agents and Tools13 min
- Threat modeling: hands-on19 min
- Threat modeling: scenarios14 min
02Volume 1: Attack Surface - Where AI Systems Break16 parts · 126 min
Understand all vulnerable layers in AI systems
- LLM Layer6 min
- RAG Layer10 min
- Agent Layer5 min
- Data Layer7 min
- API & Integration Layer6 min
- Output Layer5 min
- Model Supply Chain6 min
- Multimodal Attacks4 min
- Memory & Session Risks6 min
- Cross-Layer Thinking11 min
- Agent, Protocol & Computer-Use Surface8 min
- Reasoning Models & Hidden Context8 min
- Voice, Video & Deepfake Surface8 min
- Embedding & Vector Store Exposure8 min
- Attack surface: hands-on15 min
- Attack surface: scenarios13 min
03Volume 2: Attack Techniques - How Systems Get Broken14 parts · 135 min
Develop attacker mindset and understand real exploit strategies
- Prompt Injection Attacks9 min
- Jailbreaking Techniques7 min
- Data Exfiltration Attacks9 min
- RAG-Specific Attacks11 min
- Agent Exploitation10 min
- Cost & Resource Attacks7 min
- Model Inversion Attacks6 min
- Adversarial Inputs7 min
- Backdoors & Sleeper Attacks7 min
- Modern Jailbreak Families11 min
- Hidden-Channel Injection & Exfiltration10 min
- AI Coding & Package Attacks10 min
- Attack techniques: hands-on17 min
- Attack techniques: scenarios14 min
04Volume 3: Defense & Guardrails - Securing AI Systems15 parts · 139 min
Learn practical defense strategies and mitigation techniques
- Input Security7 min
- Output Guardrails9 min
- RAG Security8 min
- Agent Security8 min
- Access Control5 min
- Monitoring & Detection6 min
- Secure Prompt Engineering7 min
- Secrets Management4 min
- Model Protection6 min
- Defense-in-Depth11 min
- Architectural Defenses Against Prompt Injection14 min
- Guardrail Placement & Layering12 min
- Egress & Action Control12 min
- Defense and guardrails: hands-on17 min
- Defense and guardrails: scenarios13 min
05Volume 4: Secure AI System Design - Production Systems12 parts · 114 min
Design secure, scalable AI systems end-to-end
- Secure RAG Architecture10 min
- Secure Agent Systems8 min
- Reliability & Safety Engineering9 min
- Red Teaming & Testing7 min
- Model Versioning & Rollback6 min
- CI/CD Security for AI7 min
- Multi-Tenant Security6 min
- Trade-offs in Secure Design10 min
- AI Supply Chain Security11 min
- Zero Trust for AI Services11 min
- Secure system design: hands-on16 min
- Secure system design: scenarios13 min
06Volume 5: Governance, Ethics & Compliance14 parts · 123 min
Build responsible and compliant AI systems
- AI Governance Fundamentals8 min
- Risk Management9 min
- Bias & Fairness8 min
- Explainability6 min
- Data Governance5 min
- Third-Party Risk6 min
- Incident Response for AI7 min
- Human-in-the-Loop Systems5 min
- Responsible Deployment7 min
- Security Frameworks in Practice11 min
- Regulation & Compliance Timelines11 min
- Enterprise AI Policy & Vendor Controls11 min
- Governance and compliance: hands-on17 min
- Governance and compliance: scenarios12 min
07Volume 6: Real-World Use Cases6 parts · 116 min
Apply security and governance concepts in real systems
- Use Case 1: Healthcare AI Assistant24 min
- Use Case 2: Enterprise RAG System18 min
- Use Case 3: Autonomous Agent System18 min
- Use Case 4: AI Coding Agent20 min
- Use Case 5: Voice Customer Support Agent20 min
- Real-world use cases: hands-on16 min
08Volume 7: Agent & Protocol Security - Securing Agents at Scale9 parts · 120 min
Secure agents, MCP and A2A integrations, agent identity, computer-use and coding agents
- Agent Threat Model12 min
- MCP Security14 min
- A2A & Multi-Agent Trust9 min
- Agent Identity & Authorization13 min
- Computer-Use & Browser Agents10 min
- Coding Agents & CI Agents13 min
- Memory & Skill/Plugin Ecosystems10 min
- Agent and protocol security: hands-on24 min
- Agent and protocol security: scenarios15 min
09Volume 8: Privacy & Data Protection - Privacy Engineering for LLM Systems7 parts · 81 min
Protect personal data across training, retrieval, logs and vendors
- Memorization & Inference Attacks10 min
- Privacy-Preserving Training11 min
- Deletion, Unlearning & Data Subject Rights11 min
- PII Handling in LLM Pipelines11 min
- Data Residency, Retention & Shadow AI9 min
- Privacy and data protection: hands-on17 min
- Privacy and data protection: scenarios12 min
10Volume 9: Red Teaming & Guardrail Engineering - Testing and Tuning Defenses8 parts · 86 min
Run red team programs, measure security with statistics, gate releases and tune guardrails
- Red Team Program Design10 min
- Automated Red Teaming Tools10 min
- Measuring Security9 min
- Security Evals in CI-CD7 min
- Guardrail Evaluation & Tuning10 min
- Agent & Multi-Turn Red Teaming8 min
- Red teaming: hands-on21 min
- Red teaming: scenarios11 min
11Volume 10: Incident Case Studies - Learning from Real Incidents7 parts · 79 min
Analyze documented AI security incidents: root cause, missing controls, detection and postmortems
- Data Exfiltration Incidents12 min
- Agent Action Incidents11 min
- Supply Chain Incidents12 min
- Fraud & Liability Incidents7 min
- Writing the Postmortem7 min
- Incident case studies: hands-on17 min
- Incident case studies: scenarios13 min
Reviews
No reviews yet. The first comes from a reader who finishes the kit.
to review this kit once you have finished it.
Next to this one
RAG60 parts · 8h 33m · ₹299MLOps Engineer (Classical)14 parts · 20h 36m · ₹499LLMOps63 parts · 12h 23m · ₹299Healthcare RAG and Agents24 parts · 7h 44m · ₹199DSA for AI Engineers46 parts · 5h 15m · ₹249Data Scientist to GenAI100 parts · 28h 21m · ₹499AI System Design25 parts · 13h 1m · ₹299AI-GenAI Engineer108 parts · 28h 45m · ₹749AI Product Manager48 parts · 19h 53m · ₹4AI Agents66 parts · 11h 41m · ₹29Finance RAG and Agents24 parts · 8h 0m · ₹199
Forward Deployed Engineering25 parts · 15h 38m · ₹349