Understand how modern AI agents reason, plan, use tools and execute complex workflows.

Key Topics

Generative AI, LLM applications and autonomous AI agents
Agent architecture: planner, executor, memory, tools and feedback loops
ReAct, plan-and-execute and multi-agent orchestration patterns
State machines, graph-based workflows and long-running agents
Context engineering, prompt caching, compression and prioritisation
Model selection, routing and token, latency and cost optimisation
Agent loop control, termination conditions and reliable decision-making
Hands-on Project: Intelligent Task Planning Agent

Build an AI agent that understands user goals, creates an execution plan and selects the right tools and models.

Build and test practical AI agents that can perform real-world enterprise tasks securely and reliably.

Key Topics

Agent development environment and framework-free agent loops
System prompts, instruction hierarchies and advanced prompting techniques
Structured outputs, JSON Schema, function calling and tool execution
Session management, conversation state and short- and long-term memory
Human-in-the-loop approvals, handoffs and escalation workflows
Fallbacks, retries, timeouts, budget controls and failure recovery
OpenAI Agents SDK, guardrails and enterprise use cases across business functions
Hands-on Project: Enterprise Operations Agent

Build an AI agent that understands business requests, retrieves information, uses enterprise tools and seeks approval before sensitive actions.

Build context-aware AI agents that securely access enterprise knowledge, databases, applications and external tools.

Key Topics

Short-term, long-term, episodic and semantic agent memory
Cross-session memory, summarisation, retrieval and tenant-level isolation
Advanced RAG with chunking, embeddings, vector databases and hybrid search
Query rewriting, reranking, multi-hop retrieval, Graph RAG and citations
Tool engineering for APIs, databases, authentication and transactional actions
Secure tool execution with validation, approvals, rollback and sandboxing
MCP architecture, custom MCP servers and multi-server agent integration
Hands-on Project: Enterprise Knowledge & Action Agent

Build a citation-enabled RAG agent that connects to enterprise documents, databases and APIs through MCP and performs controlled actions.

Build stateful and multi-agent workflows that coordinate specialised AI agents to solve complex business problems.

Key Topics

LangGraph nodes, edges, state management and conditional routing
Checkpointing, persistence, human approvals and resumable execution
Parallel workflows, subgraphs, fault recovery and long-running agents
Task decomposition, specialised agents, supervisors and agent handoffs
Multi-agent memory, communication, conflict resolution and loop control
LangGraph, CrewAI, OpenAI Agents SDK, Microsoft Agent Framework and Google ADK
A2A protocol, MCP interoperability and automation using n8n and Make.com
Hands-on Project: Multi-Agent Research & Decision System

Build research, analysis, validation and reporting agents coordinated through LangGraph and enterprise agent frameworks.

Evaluate, secure and monitor AI agents to ensure reliable, compliant and production-ready performance.

Key Topics

Agent evaluation metrics, golden datasets and task-completion testing
Tool accuracy, retrieval quality, groundedness and citation evaluation
LLM-as-a-judge, regression testing, adversarial testing and trace grading
Prompt injection, data leakage, excessive agency and privilege escalation
Guardrails, content safety, structured validation and Responsible AI
Least-privilege access, approval gates, audit logs and kill switches
Langfuse, OpenTelemetry, tracing, cost monitoring and failure analysis
Hands-on Project: Agent Evaluation & Security Framework

Build an evaluation suite, conduct adversarial testing, implement guardrails and monitor agent performance using Langfuse.

Deploy, scale and manage secure, reliable and production-ready AI agents.

Key Topics

FastAPI-based agent services, REST APIs, webhooks and streaming
Docker, cloud, on-premises and serverless deployment
Persistent state, queues, caching and background workflows
Authentication, role-based access and secret management
CI/CD, automated testing, versioning and rollback strategies
Scaling, concurrency, latency and infrastructure cost optimisation
AgentOps governance, production monitoring and incident management
Hands-on Project: Production Deployment of an AI Agent

Deploy a containerised AI agent with authentication, persistent state, CI/CD, observability, evaluation and rollback support.

Apply everything learned throughout the program to build and present an end-to-end, enterprise-ready Agentic AI solution for a real-world business problem.

Capstone Deliverables

Business problem, user journeys and system architecture
Model selection, routing, tool schemas and context design
Enterprise RAG, MCP integration and multi-agent workflows
Human approvals, guardrails, access controls and security design
Evaluation datasets, quality metrics and failure-recovery strategy
FastAPI application, Docker deployment and CI/CD pipeline
Observability, cost analysis, business ROI and final project demonstration