AI-Powered Cybersecurity Training Course

Part 1: Understanding and Using AI

• • Module 1: Introduction to AI and AI in Cybersecurity
    • AI and its Evolution
    • Components of an AI System
    • AI vs ML vs Data Science
    • Algorithm vs Model
    • Types of Learning – Supervised, Unsupervised, Semi-Supervised, Reinforcement, Federated
    • Types Of Models
    • AI Use Cases in Cybersecurity
    • Risks of AI
  • Module 2: Python Basics for Using AI Frameworks
    • Python Fundamentals for AI
    • Lab Setup: Linux VM, Using Jupyter Notebook
    • Python Libraries for Data Analysis and Visualisation – Pandas, NumPy, Matplotlib
    • GenAI Platforms: Google Colab, Hugging Face, Ollama, LMStudio, OpenAI Platform, Google AI Studio, Anthropic, Groq
    • AI Model Development Lifecycle
    • Types of Data and using them for AI Model development
    • Secure and Responsible use of Data in AI models and Applications
  • Module 3: Using AI for Cyber Offense
    • Using AI for:
      • Reconnaissance
      • Vulnerability Scanning
      • System Hacking
    • Hands On: ShellGPT, Open-Source Models for Offensive Security
    • Hands On: Generating Network Attacks Using AI
    • Hands On: Collecting Network Attack Traffic Data: Packets and Logs.
  • Module 4: Building ML-based Security Controls using Generative AI
    • Data Collection
    • Data Processing (Including Feature Extraction and Normalization)
    • Model Training
    • TEVV: Test, Evaluation, Verification, Validation, and Fine Tuning
    • Model Deployment, Model Monitoring and Retraining
    • Hands On: Using Gen AI tools for Building AI models for Network Security
    • Hands On: Building Anomaly Detector using Unsupervised Learning using Gen AI Tools
    • User and Entity Behaviour Analytics (UEBA)
    • Hands On: Building Keystroke Behaviour Analysis Detector
  • Module 5: Natural Language Processing (NLP)
    • Text Processing for NLP: Tokenization, Stop words removal
    • NLP Feature Engineering Types
    • Hands On: Converting Emails and Logs to Vectors
    • Hands On: Building a Phishing Mail Detector
  • Module 6: Neural Networks and Deep Learning for Cybersecurity
    • Perceptron
    • Multi-Layer Perceptron (MLP)
    • Convolutional Neural Networks (CNN)
    • Recurrent Neural Networks (RNN)
    • How Deep Learning enables Building Models for Complex Cybersecurity Data
    • Hands On: Building Malware Detectors Using CNN
  • Module 7: Generative AI and LLMs
    • Generative Adversarial Networks
    • Transformer Architecture
    • Prompt Engineering Techniques and System Prompts
    • Foundational model and fine-tuned model
    • Retrieval Augmented Generation (RAG)
    • Hands On: Building a Gen AI-Based Custom Chatbot
    • Hands On: Using Gen AI for Summarizing Threat Intelligence and Vulnerability Reports
  • Module 8: SIEM-IDS-Integration, Deployment, and Monitoring of AI Models
    • Ingesting IDS Logs into SIEM (ELK Stack)
    • Data Parsing using AI
    • Setting up the Data and Model Pipelines
    • Hands On: FastAPI for Deploying and Serving Models
    • Monitoring and Observability
    • Dealing with Data Lag and Model Drift
    • Secure Retraining Mechanisms
    • Hands On: AI Observability with OpenLLMetry + Phoenix
  • Module 9: Agentic AI for Security Operations
    • Components To Build Agentic AI
    • Agentic AI: Autonomous agents, Reasoning, Action: tool use, and multi-agent orchestration frameworks
    • Model Context Protocol (MCP): architecture, use cases, and security implications
    • Demo: Using Agentic AI for generating Triage Reports
    • Agent topology design principles.

Part 2: Attacking, Governing, and Securing AI

• Module 10: Adversarial Attacks on AI
  • Poisoning: Data and Model
  • Setting up Backdoor with Trigger-Based Poisoning
  • Evasion
  • Evading AI-Based Detectors
  • Theft: Data and Model
  • Inversion and Inference Attacks
  • OWASP Machine Learning Security Top 10
  • OWASP Top 10 for Large Language Model Applications
  • OWASP Top 10 for Agentic Applications
  • Hands On: Honeytokens in prompts and vector stores
  • Hands On: Automated LLM pentesting using Garak
• Module 11: Security Architecture Principles Applied to AI
  • Secure Infrastructure reference architecture for AI Workloads
  • Defense-in-depth for AI pipelines
  • Zero-trust architecture for AI services
  • Role of AI Gateway
  • Segmentation of data, model, and inference layers
  • Identity and Access Control for AI Systems
  • Hands On: Deploying AI Gateway
  • Hands On: Shadow AI Discovery using Proxy Server
• Module 12: Data Security in AI Systems
  • Data Confidentiality, Integrity, and Availability
  • The AI data leakage threat surface
  • Secure Data Collection and Ingestion
  • Data Sanitization and De-identification Techniques: Data Anonymization, Pseudo-anonymization, Minimization, Masking, Redaction
  • Secure Data Processing: Data Classification, Data Quality, and Validation
  • Dataset Watermarking
  • Privacy Preserving Learning Techniques
  • Secure Data Usage During Inference
  • Dealing with Data Lag after Model Deployment
  • Data Lineage and Provenance Tracking: Audit Logging, Versioning
  • Hands On: Using Microsoft Presidio
• Module 13: AI Governance
  • AI Ethics
  • Properties of a Responsible AI System
  • Data Governance
  • Model Governance
  • Mapping Infosec controls to NIST AI RMF functions (Govern, Map, Measure, Manage) and ISO 42001 Annex A controls
  • EU AI Act risk tiers and their implications for security controls.
  • Hands On: Governance control mapping exercise
• Module 14: Implementing Governance and Security Controls for AI
  • Adversarial Training and Testing
  • Techniques to check Sampling and Algorithmic Bias
  • Ensuring Explainability in AI Systems
  • Data and Model Versioning
  • Guardrails for LLMs: System Prompt Hardening, Prompt Firewalls (Input and Output Guards)
  • Integrating Prompt Injection Detector with LLMs
  • Proprietary and open-source guardrails: LLM-Guard, Nemo Guardrails, Guardrails-AI, LlamaGuard
  • Hands On: Deploying a pre-built Prompt Injection Detector and interpreting its alerts
  • Applying Rate Limiting and Cost Budgeting in the AI Gateway
  • Observability, Monitoring, Evaluation, and Logging Tools for LLMs
  • Hands On: Third-party AI Component Supply Chain Assessment
  • Building a Governance Evidence Pack for an AI Application
  • Threat Modelling Frameworks: MITRE ATLAS, MAESTRO Framework for Agentic AI systems
  • Security Roles, Responsibilities & Operating Model for AI