Practice Module for Architecting AI Systems

Overview

Part of	Graduate Certificate in Architecting AI Systems
Duration	15 days
Course Time
Enquiry	Please contact ask-iss@nus.edu.sg for more details.

In the Graduate Certificate in Architecting AI Systems, participants will learn to apply essential practices in Explainable and Responsible AI, AI-specific cybersecurity, Agentic AI design patterns, advanced software architecture skills, and MLSecOps/LLMSecOps to design, architect, implement and deploy AI systems that are robust, highly available, reusable, maintainable, and extensible, along with the development of relevant project artifacts.

This Graduate Certificate comprises four course modules and a practice module. The four course modules are:

Explainable & Responsible Artificial Intelligence
AI & Cybersecurity
Architecting Agentic AI Solutions
Integrating and Deploying AI Solutions

The sections below provide more details about the practice module.

To be awarded the graduate certificate, participants must demonstrate competency in all four course modules and undertake and pass the practice module. The graduate certificate, once awarded, may also be used as one component in a stackable Master of Technology in Software Engineering (MTech SE) and Master of Technology in Artificial Intelligence Systems (MTech AIS).

Objectives

This practice module is designed to achieve two key objectives:

Firstly, to provide participants with exposure to real-world problems where they can apply the skills acquired in individual course modules to develop practical solutions. This will be carried out under supervision, allowing students to seek expert guidance and support when needed.
Secondly, to enable participants to demonstrate their proficiency across the range of skills acquired throughout the course modules, thereby qualifying for certification at the Certificate level.

Intended Audience

This Practice module is targeted at participants who wish to complete the certification process for the Graduate Certificate in Architecting AI Systems.

Prerequisites

Participants must have successfully completed (or in the progress of completing or have been exempted from) all the four course modules of the Graduate Certificate in Architecting AI Systems.

Components

There is one part to the practice module.

1. The participants will need to undertake a practice project to gain practical experience and demonstrate their understanding and mastery of the skills taught in the four component courses. The practice project will require each participant to expend an estimated 15 man-days of effort. These days are not expected to be contiguous and may stretch over many weeks. These practice projects are conducted in teams with significant individual elements.

The overall grade for the participants will be based on the practice project.

Typical example of a project to be undertaken

Problem Description

CARE-AI is a telemedicine enhancement initiative for a regional virtual clinic that aims to streamline pre-consult triage, improve patient education, and reduce repetitive workload for healthcare staff. The proposed solution involves a multi-agent system, where specialized AI agents collaborate to support patients before and after online consultations.

Key agent roles and system features include:

Triage Agent – Collects symptoms and classifies urgency level
Preparation Agent – Guides patients in preparing for their consultation (e.g., blood pressure logs, medication list)
Education Agent – Answers common questions using vector-based medical information retrieval
Follow-up Agent – Provides post-consultation advice and reminders (e.g., medication, lifestyle guidance)
Escalation Agent – Monitors conversations and flags safety-critical cases for human intervention

Agents may coordinate using shared memory or messaging. Each agent is expected to demonstrate autonomy through reasoning, planning, and tool use. The system should also maintain logs to ensure traceability and explainability of agent decisions.

The project may use simulated data and is not expected to be production-ready. The emphasis is on demonstrating agentic behaviour, explainable and responsible AI practices, AI-specific security considerations, and sound engineering principles including thoughtful architecture, modular design, and MLSecOps/LLMSecOps practices for integration, testing, and monitoring.

Deliverables and Success Criteria

Presentation slides clearly outlining the problem context, agent roles, system overview, and demo walkthrough.
System architecture document providing a high-level view of the overall system. This should cover the logical and physical architecture, infrastructure setup, service deployment strategy, integration points, data flow across components, and justification of architectural styles and technologies used.
Agent design documentation describing the internal logic and behaviour of each agent. This includes the reasoning patterns used, planning loops, memory mechanisms, tool integrations, and how autonomy is achieved. Also specify prompt patterns, fallback strategies, and inter-agent communication protocols.
Explainable and Responsible AI report explaining for example, how fairness, bias mitigation, explainability, and governance principles were considered and applied.
AI security risk register identifying risks resulting from potential vulnerabilities (e.g., prompt injection, hallucination), including AI agent-related risks, mitigation strategies and appropriate security controls.
MLSecOps / LLMSecOps pipeline design specifying tools and workflows for CI/CD, automated testing (including AI security tests), model versioning, monitoring, and logging.
Well-structured source code repository showcasing modular agent implementation with clear separation of logic and well-documented code.
Testing artifacts including unit tests for agent behaviour, end-to-end flow verification and relevant AI security tests.
Simple UI prototype (e.g., web-based or form-based) to demonstrate user interaction and multi-agent orchestration.

Application (For Stackable Students)

	Semester 1 (Jul to Nov)	Semester 2 (Jan to May)
Application*	15 Apr to 15 Jun	15 Oct to 15 Dec
Payment Deadline	30 Jun	31 Dec
Briefing	First two weeks of Jul	First two weeks of Jan

* Eligible participants will be contacted 1 week after application closure.

Note:

Participants are only allowed to take the practice module after completing all courses in the Grad Cert.
Participants who wish to take the practice module concurrently in the same semester with the courses in the same Grad Cert must write to ask-iss@nus.edu.sg citing reasons by the application deadline. Email requests received after the deadline will not be considered. Requests will be reviewed after the deadline and approved on a case-by-case basis.
Participants who miss the application window will have to apply for the practice module in the next semester.
Participants who do not attend the briefing will be withdrawn from the practice module.

Apply Here

Course Resources

Develop your Career in the Following
Training Roadmap(s)

Please click on the discipline(s) to view the training roadmap of related courses to assess your training needs and goals.