GenAI and Agentic AI for Software Engineering
Comprehensive hands-on program on integrating AI agents into software development. Learn to build software faster, more stable, and more scalable by using AI as part of the SDLC, not just as a tool.
GenAI and Agentic AI for Software Engineering
Overview
AI in software development is no longer an experiment — it’s a new engineering standard that is transforming coding, review, testing, and system architecture today. AI won’t take your job. But engineers who don’t integrate AI into their workflow will gradually lose their competitive edge.
This program represents the next step in software engineering: AI-assisted and agent-driven development. You will learn to build reproducible workflows, work with specifications and agents, and design systems fully integrated into the SDLC. By the end, you will think like an AI-augmented engineer, with increased speed and quality in your projects regardless of domain.
what you will learn
requirements
Programme
Workshop 1: Introduction & Hello World with AI
- Introduction to AI coding assistants (GitHub Copilot, Claude Code, Cursor)
- Tool comparison and use cases
- Setting up development environment
- Building a simple REST API with AI assistance
- Comparing CLI vs IDE-based workflows
- Build a simple Node.js web application with a "Hello World" endpoint using:
- GitHub CopilotCLI workflow
- Claude Code workflow
- Compare approaches and results
Workshop 2: Foundations - Context, Model, Prompt Framework
- The Context-Model-Prompt triad
- Context management best practices
- Model selection for different tasks
- Prompt engineering fundamentals
- Common mistakes and how to avoid them
- Build a banking transaction parser that:
- Handles multiple formats (CSV, JSON, XML)
- Detects potential fraud patterns
- Demonstrates proper context management
- Tests with different models (small vs reasoning models)
Workshop 3: Specification-Driven Development
- Specification-driven development principles
- Specification template structure
- Agent configuration files (agents.md, cursorrules, claude.md)
- API-first development approach
- Contract testing and validation
- Build a KYC (Know Your Customer) API following specification-driven approach:
- Review and understand specification document
- Create agent configuration file
- Implement API using AI assistance with specification
- Validate implementation against specification
Workshop 4: Model Benchmarking & Security, Skills
- Industrial performance benchmarks (SWE-bench, GitHub Copilot rankings)
- Model selection strategies
- Cost vs quality trade-offs
- Reasoning models vs fast models
- Performance metrics in real-world scenarios
- AI usage security
- Adding skills for AI agents
- Explore benchmark results and leaderboards
- Test same task with different models
- Compare results and performance
- Analyze cost implications
Workshop 5: AI Developer Workflows (ADW) - Multi-Agent Systems
- AI Developer Workflows (ADW) concepts
- Architect-Editor pattern
- Multi-agent orchestration
- Agent types: Test Runner, Security Auditor, Documentation Generator, etc.
- Integration patterns (side-car, internal, API, service)
- Review fraud detection application code
- Select 2-3 agents to build
- Create agent specifications
- Present agent selection rationale
- Implement selected agents
- Integrate agents with fraud detection system
- Demonstrate multi-agent collaboration
- Build additional features using agents
Workshop 6: PR Review & Automation
- AI-assisted PR review workflows
- Tool comparison: Cursor vs Claude Code vs Codex in code review
- CI/CD integration strategies
- Configuration of an automated code review system
- Compare tool outputs and recommendations
- Write comprehensive PR review comments
Workshop 7: Model Context Protocol (MCP)
- MCP architecture and benefits
- Official MCP servers (Filesystem, Git, Memory, AWS, etc.)
- Enterprise MCP servers (GitHub, Azure, PostgreSQL, etc.)
- Multi-server orchestration
- Custom server development
- Configure multiple MCP servers (GitHub, AWS, PostgreSQL, etc.)
- Build MCP-powered data assistant
- Create workflows using multiple servers
- Demonstrate multi-server orchestration
Capstone Project Presentations
Applying learned techniques and using course materials, tools and frameworks, students are expected to develop a project. They will work on building a complete distributed system using interconnected AI agents that process tasks through a full pipeline. Teams may either use the proposed project template or choose a topic of their own interest.
Successful preparation and completion of the final capstone project is mandatory for receiving course completion certificate.
instructor
Oleksii Popov
VP of Engineering at GenAI.Works, with 15+ years of experience in engineering leadership and solution architecture. Experienced in leading global teams and large-scale projects, from software engineering to Head of Engineering roles. Over 8 years in solution architecture at EPAM, Ciklum, and Customertimes, designing scalable, cloud-native solutions and implementing innovative technologies.
Benefits
Hands-on — workshops, exercises, and production-oriented case studies
Tool-first — work with real tools and protocols, not just theory
Systematic — from basic AI principles to multi-agent systems in SDLC
Expert-led — learn from a practitioner in AI engineering productivity who shares what really works
Practical deliverables — reproducible workflows, specifications, configurations, and a final Capstone Project
who it's for
Software Engineers (Back/Full-Stack/SDET/Automation QA)
Tech Leads / Engineering Managers
Developers who strive for growth and want to integrate AI tools into their software development workflow
faq
May I apply for the program if I don't use / know Python / JS?
Yes. The main focus of the course is on tools, workflows, and principles (Context → Model → Prompt, specifications, MCP, agents) that are independent of any specific programming language. Python/JavaScript are used only for demonstration purposes.
Learn more about the SET University program
