The goal of the course is to equip computer science students with the core competencies of modern software engineering while introducing them to the principles and practices of building AI agent systems.

Lectures

1. Software engineering goals, lifecycle models, team roles, and an introduction to AI agents. Defining AI agents, distinguishing them from simple chatbots, and explaining autonomy, tool use, memory, and action as core agent properties. 2. Stakeholder analysis, functional and non-functional requirements, user stories, acceptance criteria, and use case modeling. Showing how requirements for AI assistants and agentic workflows differ from requirements for conventional business software. 3. UML as a communication and specification language. Focus on use-case diagrams, class diagrams, object diagrams, database diagrams and package-level structure. Preparing students for Mermaid-based class diagram modeling in the labs. 4. Activity diagrams, state machines, sequence diagrams, and interaction modeling. Applying them to traditional systems and to AI agent workflows such as observe–think–act loops and tool invocation chains. 5. Architectural decomposition, interfaces, coupling and cohesion, and basic architectural styles. Agent architecture: planning, tool use, memory, orchestration, and multi-agent collaboration as an application architecture topic. 6. Conceptual data modeling, relational design, normalization (recapitulation), and mapping from model to database. Showing how persistence supports business applications and agent systems that require memory, logs, or task state. 7. Coding standards, code review, branching strategies, Git-based collaboration, and traceability from model to code. How agent-generated code requires review, testing, and version control like human-written code. 8. Scrum, iteration planning, backlog management, and lightweight project governance. Short overview of other approaches: Extreme Programming, Test-Driven Development, etc. 9. SOLID principles and a curated set of design patterns: factory, adapter, decorator, proxy, facade, observer, command, iterator, and MVC. Examples that support extensibility, dependency control, and agent tool abstractions. 10. Midterm. 11. Unit, integration, system, and acceptance testing, Test-Driven Development. AI-agent evaluation basics: task success, error analysis, trace inspection, and measurement of goal completion reliability. 12. Agent design patterns such as reflection, tool use, planning, memory, and multi-agent coordination, and comparing them with classical software engineering practices. Documentation, testing, and presentation practices for agentic software projects. 13. Make-up/retake midterm.

Labs

1. Introducing the semester project, team roles, and a Git-based collaboration workflow. Creating a Git repository, defining a branching strategy, and setting up an issue tracker. Short introduction to AI agents: students observe a simple agent demo, distinguish it from a chatbot, and identify autonomy, tool use, memory, and action as core properties. 2. Guiding students in eliciting requirements for the project, performing stakeholder analysis, and writing user stories with acceptance criteria. Creating use case diagrams in Mermaid and rendering them using mermaid-drawing.com. Showing how requirements for AI assistants and agentic workflows differ from requirements for conventional business software. 3. Use-case, class, object, database, and package diagram syntax in Mermaid and rendering them online using mermaid-drawing.com. Modeling the main domain classes, their relationships, and the database schema for the project. Establishing Mermaid as the default diagramming tool for UML structural modeling. 4. Activity, state machine, and sequence diagram syntax in Mermaid. Modeling activity diagrams, state machines, and sequence diagrams in Mermaid. Including at least one sequence diagram with an AI agent calling a tool or external service. Validating diagrams by rendering and peer review, and applying behavioral models to both traditional systems and agent workflows such as observe–think–act loops. 5. Creating an architectural decomposition of the system using Mermaid diagrams (packages, components, and interfaces). Defining coupling and cohesion for key modules and selecting an architectural style. Designing a simple agent architecture with planning, tool use, memory, orchestration, and optionally multi-agent collaboration as an application architecture component. 6. Creating a conceptual data model, performing relational design and normalization, and mapping the model to a relational schema using Mermaid database diagrams. Showing how persistence supports business applications and agent systems that require memory, logs, or task state. Optionally generating a database schema from the model as a demo. 7. Implementing a small feature following coding standards, performing a peer code review, and using Git branching and merging in a team context. Creating traceability links between UML diagrams and code. Discussing how agent-generated code requires review, testing, and version control like human-written code. 8. Running a short sprint-planning exercise in which students define a backlog, estimate tasks, and assign work. Practicing Scrum roles, events, and artifacts, and tracking progress during a simulated iteration. Including a short overview of Extreme Programming and Test-Driven Development as complementary agile practices. 9. Refactoring a small code base to improve responsibility boundaries and dependency direction according to SOLID principles. Implementing one or two selected design patterns (factory, adapter, decorator, proxy, facade, observer, command, iterator, or MVC) and discussing how the pattern supports extensibility, dependency control, and agent tool abstractions. 10. Writing unit, integration, and system tests using Test-Driven Development for a small feature. Performing acceptance testing and evaluating an AI agent component by measuring task success, analyzing errors, inspecting traces, and assessing goal completion reliability. Connecting classical testing practices with AI-agent evaluation basics. 11. Integrating the AI agent component with the main software project, applying agent design patterns such as reflection, tool use, planning, memory, and multi-agent coordination, and documenting the system using Mermaid diagrams. 12. Semester project presentation. 13. Project presentation make-up/retake.

Lecture notes - available updated after each lecture in the Moodle course of the subject.