Advanced AI Agents Masterclass: Engineering, Security, and Governance for Production Systems Training Course

Day 1

Structure of Contemporary AI Agents

Moving beyond chatbots to agents as autonomous reasoning and action systems

Reactive, proactive, hybrid, and goal-driven agent paradigms

Essential components: perception, planning, memory, tool utilization, and action

Trade-offs in designing single-agent versus multi-agent systems

Agent Frameworks and the Modern Technology Stack

Evaluation of LangChain, LlamaIndex, AutoGen, and CrewAI, including their respective trade-offs

Comparison with classical frameworks like JADE and SPADE

Selecting a framework based on production requirements

Tool invocation, function calling, and structured output generation

Practical exercise: Creating a single Python agent equipped with tool calling capabilities

Multi-Agent System Architectures

Designing centralized, decentralized, hybrid, and layered MAS structures

FIPA ACL, message passing, and their modern counterparts

Coordination patterns: planning, negotiation, and synchronization

Emergent behaviors and self-organization within agent populations

Decision-Making and Learning in Agents

Applying game theory to cooperative and competitive agent interactions

Reinforcement learning in multi-agent environments

Transfer learning and knowledge sharing between agents

Conflict resolution and trust mechanisms among coordinating agents

Day 2

Foundations of Multi-Modality for Agents

Multi-modal AI as a unified workflow spanning text, image, speech, and video

Leading multi-modal models: GPT-4 Vision, Gemini, Claude, and Whisper

Fusion techniques for integrating modalities within an agent's reasoning loop

Trade-offs regarding latency, cost, and accuracy in multi-modal pipelines

Constructing the Perception Layer

Image processing for agents: classification, captioning, and object detection

Speech recognition using Whisper ASR and streaming transcription

Text-to-speech synthesis and natural voice interactions

Linking perception outputs to LLM-driven reasoning and tool selection

Hands-On - Developing a Multi-Modal Agent in Python

Defining the agent's tasks, context window, and tool inventory

Connecting GPT-4 Vision and Whisper APIs end-to-end

Implementing memory, state management, and conversation handling

Integrating tool calls that safely produce real-world side effects

Hands-On - Orchestrating a Multi-Agent System

Composing specialized agents using AutoGen or CrewAI

Defining roles, responsibilities, and inter-agent communication protocols

Resource allocation and coordination in simulated environments

Logging agent reasoning, tool calls, and decisions for inspection and auditing

Day 3

Threat Surface of Production AI Agents

Unique vulnerabilities of agentic AI compared to traditional software

Attack surfaces: data, model, prompt, tool, output, and interface layers

Threat modeling for agent-based systems with autonomous tool usage

Contrasting AI cybersecurity practices with traditional cybersecurity

Adversarial Attacks Practical

Adversarial examples and perturbation methods: FGSM, PGD, DeepFool

White-box versus black-box attack scenarios

Model inversion and membership inference attacks

Data poisoning and backdoor injection during the training phase

Prompt injection, jailbreaking, and tool misuse in LLM-based agents

Defensive Strategies and Model Hardening

Adversarial training and data augmentation strategies

Defensive distillation and other robustness techniques

Input preprocessing, gradient masking, and regularization

Differential privacy, noise injection, and privacy budgets

Federated learning and secure aggregation for distributed training

Hands-On with the Adversarial Robustness Toolbox

Simulating attacks against the multi-modal agent developed on Day 2

Measuring robustness under perturbation and quantifying performance degradation

Iteratively applying defenses and re-evaluating attack success rates

Stress-testing tool-call pathways and prompt injection vectors

Day 4

Risk Management Frameworks for AI

NIST AI Risk Management Framework: govern, map, measure, and manage

ISO/IEC 42001 and emerging AI-specific standards

Aligning AI risks with existing enterprise GRC frameworks

Requirements for AI accountability, auditability, and documentation

Regulatory Compliance for Agentic Systems

EU AI Act: risk tiers, prohibited uses, and obligations for high-risk systems

Implications of GDPR and CCPA for agent data pipelines

U.S. Executive Order on Safe, Secure, and Trustworthy AI

Sector-specific guidance for finance, healthcare, and public services

Third-party risks and supplier AI tool usage

Ethics, Bias, and Explainability

Detecting and mitigating bias across agent perception and reasoning

Explainability and transparency as security-relevant properties

Fairness, downstream harm, and responsible deployment

Designing inclusive and auditable agent behavior

Production Deployment, Monitoring, and Incident Response

Secure deployment patterns for single and multi-agent systems

Continuous monitoring for drift, anomalies, and abuse

Logging, audit trails, and forensic readiness for agent actions

AI security incident response playbooks and recovery procedures

Case studies on real-world AI breaches and key takeaways

Capstone and Synthesis

Reviewing the multi-modal multi-agent system developed throughout the course

End-to-end pipeline review: design, construction, security, governance, and deployment

Self-assessment of the system against NIST AI RMF functions

Future outlook on emerging trends in agentic AI and AI security

Summary and Next Steps