AI Software Engineering Fundamentals
or software development with AI
How to work effectively and deliberately with coding agents: Claude Code and Codex on real projects, project context, skills, MCP integrations, multi-agent workflows and autonomous systems wired into CI.
Agentic coding isn't slowing down: new frontier models (Claude, GPT), coding agents that work on their own for hours, a maturing MCP ecosystem, the spread of skills and the arrival of headless agents have noticeably redrawn the map in just six months.
That's why AI Software Engineering Fundamentals is back: built on the timeless foundations of the first edition, but with an updated model overview, a reworked tooling chapter and the latest multi-agent and autonomous workflow patterns. We keep updating the syllabus, so the latest tools and patterns are always part of the material.
Models don't just assist anymore, they drive: they plan, use tools, work on their own for hours and solve increasingly complex software engineering tasks without human intervention. If you learn to work this way, you can be orders of magnitude more effective. But you have to understand what you're doing and why: writing the code is less and less the job, while engineering thinking has never been more valuable.
This hands-on course shows you how to work with these tools effectively and deliberately. We walk through how modern agentic workflows are built, show how to take real projects forward with Claude Code and Codex, how to design project context, skills and MCP integrations, and how multi-agent systems and autonomous systems wired into CI are put together.
If you feel like the world is changing around you faster than you can catch up, this course is for you!
8 modules
Syllabus
-
From LLMs to agentic thinking
- How LLMs work through an engineer's eyes: tokenization, context window, system prompt, reasoning models
- Prompt engineering vs. context engineering
- The model ecosystem
- Pricing basics
-
Agent architecture
- What separates a chatbot from an agent?
- Tool calling and structured output schemas
- Agentic loop basics: Reason → Plan → Act → Observe
- Separating planning from execution
- Breaking down tasks and error handling strategies
-
The agentic tooling landscape (IDE vs. CLI)
- Claude Code vs. Codex vs. OpenCode
- IDE-native agents: Cursor, Antigravity, Zed, VS Code extensions
- CLI-native agents and the benefits of a terminal-based workflow
- Differences in plan quality and context handling
- Choosing an agent: project size, level of control, auditability
-
Context engineering and project memory design
- Project-level memory: the role of instruction files
- Instruction layering and designing rule sets
- Optimizing context density
- Getting deterministic behavior out of an emergent model
-
Agent skills: structured domain intelligence
- What is a skill, and how is it different from a tool or MCP?
- Skills as behavior modules: workflow, constraints, activation conditions
- Progressive disclosure and reducing context load
- Types of coding skills: refactor, TDD, migration, security review
-
MCP: the capability layer and external integrations
- What is the Model Context Protocol, and why does it matter?
- The Host–Client–Server architecture
- Connecting tools: databases, GitHub, communication channels
- Capability sandboxing and permission management
-
Multi-agent architecture and skill composition
- Subagents and delegation patterns
- Role-based agents (Planner, Builder, Reviewer)
- Quality gates and automated verification
- Reviewer loops and self-correcting agent systems
-
Autonomous coding systems and production governance
- Headless agents and CI/CD integration (GitHub Actions)
- Continuous agent loops
- Agent sandboxing
- Security risks (prompt injection, tool misuse, over-permission)
Interested in this course?
Drop us a line and we'll figure out how to tailor the course to your team.