Module Specifications
Academic Year 2026 - 2027
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Description This module provides students with both a conceptual and practical grounding in engineering AI agents based on large language models. It introduces the foundational concepts underpinning prompt engineering and LLM-based agents, including agent design, orchestration, and evaluation, helping students develop an intuitive understanding of how these systems work and how to use them effectively. The module focuses on contemporary practices including basic and advanced prompt engineering, retrieval-augmented generation (RAG), reasoning and planning, memory and context management, tool calling, agent orchestration, and multi-agent collaboration. Students will explore techniques for controlling and evaluating agent behaviour, including systematic evaluation of prompts, inputs, models, and outputs, automated evaluation frameworks, bias, privacy, and safety considerations, and agent benchmarking. Ethical, legal, and societal implications of deploying generative AI systems will be emphasised throughout the module, enabling students to engage critically with the risks, limitations, and broader impact of these technologies. The module consists of lectures and practical activities, including quizzes, exercises, and projects. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Learning Outcomes 1. LO1. Explain the core concepts and architectures of LLM-based agents, including prompt engineering, retrieval-augmented generation, memory and context management, planning and task decomposition, and tool calling. 2. LO2. Design agent solutions, including agent orchestration and multi-agent collaboration, for a given task. 3. LO3. Select and evaluate the appropriate agent frameworks, tools, and open-source implementations for applied or research-oriented use cases. 4. LO4. Evaluate agent behaviour and outputs using appropriate methodologies, including automated evaluation frameworks and benchmarking, and communicate results through a structured technical artefact or report. 5. LO5. Apply quality assurance approaches for agent systems, including robustness and reliability checks, safety considerations, and bias and privacy evaluation. 6. LO6. Integrate tools, memory, orchestration strategies, and evaluation components into an end-to-end AI agent system addressing a realistic task. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
All module information is indicative and subject to change. For further information,students are advised to refer to the University's Marks and Standards and Programme Specific Regulations at: http://www.dcu.ie/registry/examinations/index.shtml |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Indicative Content and Learning Activities
Activities ● Foundations of Generative AI and LLM-Based Agents Overview of large language models, generative AI paradigms, and the role of agents; capabilities, limitations, and system-level considerations. ● Prompt Engineering Fundamentals Basic prompt design patterns, instruction following, role prompting, few-shot and zero-shot prompting; prompt iteration and refinement. ● Advanced Prompt Engineering and Reasoning Techniques Structured prompting, reasoning strategies, self-reflection and critique patterns, and prompt optimisation techniques. ● Retrieval-Augmented Generation (RAG) Principles of retrieval-based systems; document ingestion, embedding, indexing, retrieval strategies, and integration with LLMs. ● Memory and Context Management Short-term and long-term memory for agents; context windows, summarisation, and memory architectures. ● Tool Calling, Agent Protocols, and External System Integration Design and use of tools for LLM-based agents; APIs, function calling, code execution, interaction with external services, and introduction to agent communication protocols (e.g., MCP). ● Agent Design, Planning, and Orchestration Single-agent and multi-agent systems; planning, task decomposition, collaboration, and orchestration strategies. ● Evaluation, Quality Assurance, and Responsible Use Evaluation frameworks and testing methodologies; assessment of prompts, inputs, models, and outputs; robustness analysis, failure modes, agent benchmarking, and considerations of safety, bias, privacy, and ethical use. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Indicative Reading List Books:
Articles: None | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Other Resources None | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||