AI Application: Retrieval-Augmented Generation (RAG) and Agentic Workflow

Title	AI Application: Retrieval-Augmented Generation (RAG) and Agentic Workflow
Course code	CM673-08-2025-C
Objective	This course is specifically designed for Python developers, offering an in-depth exploration of the core principles and applied techniques of large language models (LLMs). It aims to help participants move beyond the basic chatbots, equipping them with key skills to build knowledge databases, implement agentic workflows, and master Retrieval-Augmented Generation (RAG). Using practical case studies such as legal document analysis, trip planning, and financial data analysis, the course provides a hands-on learning experience. However, learners are not limited to these examples; by the end of the course, they will be able to flexibly apply the acquired techniques to a wide range of scenarios, solving real-world problems and exploring innovative applications to unlock the full potential of AI.
Content	Introduction to AI Engineering and Experimentation: Learn the distinctions between AI engineering and machine learning engineering, master the applications of Prompt Engineering, and explore the capabilities ofv LLMs through hands-on experiments Vectorized Database Foundations: Understand embedding models and vectorized databases, build a basic legal document knowledge base, and grasp practical applications of data retrieval and storage RAG Technology in Practice: Develop a Retrieval Augmented Generation (RAG) intelligent Q&A system, learn performance evaluation methods, and apply the system to legal knowledge base scenarios Agentic Workflows and Applications: Explore components of agentic workflows such as prompt chaining and parallel processing, and design multi-step task management systems for everyday scenarios Trip Planning Application Development: Combine RAG and workflow techniques to create a travel planning application capable of recommending destinations and generating itineraries Agentic Framework and Implementation: Learn agentic frameworks that are commonly used, rebuild the intelligent legal assistant system, and understand the value of modular design Low-Code Solutions and Financial Data Analysis: Use no-code/low-code tools to quickly build intelligent systems for retrieving and analyzing financial data, lowering the technical barrier Comprehensive Project: Independently design and implement a project, integrating course techniques to create innovative applications, and refine the project based on feedback
Assessment	At least 80% attendance and complete all in-class exercises and assessment.
Target audience	Individuals interested in artificial intelligence (AI) and its applications, including professionals aiming to implement intelligent solutions in specific domains, as well as technical developers seeking to delve into AI engineering practices.
Prerequisite	Basic understanding of Python (or completed CPTTM course CM540 'Introduction to Python programming'), and with a logical thinking and problem solving mindset.
Class size	18
Instructor	CPTTM Appointed Instructor
Handout	All printing material, classroom, instructor, course evaluation report & students' result report provided by CPTTM
Instruction language	Cantonese (supplemented with English)
Handout language	Handouts partly in Chinese and partly in English
Duration	24 hours in 8 sessions
Schedule	10:00-13:00, from Aug 9, 2025 to Aug 31, 2025 every Saturday, Sunday.
Fee	MOP2,980
Venue	Cyber-Lab (Rua Comandante Mata Oliveira, Ed. Associacao Industrial, 3-andar Macau)
Certificate	Certificate of Completion issued by CPTTM (with at least 80% attendance and passed the assessment)
PDAC code	Approved Course under the SAR Government "Continuing Education Development Program", Code: 2504140078-0