Software Architecture for AI.

The better AI gets at writing code, the more your value shifts to architectural skills. In this workshop, you’ll learn essential software architecture principles, practices, and patterns for the age of AI-assisted software engineering.

Introduction

Coding skill alone is not enough to get or keep a job in an AI world, no matter how good you are. AI can code, but it can't design. That requires judgment, and that's what keeps you ahead.

Good architecture makes your system easier to build, maintain, and evolve. It also shapes how your team works — a team fighting its architecture can't move fast, and agility becomes impossible.

Architecture is particularly important when you integrate AI into your workflow. Even Spotify, which claims 100% of its new code is machine-generated, doesn't entrust its LLM with architectural decisions. A poorly designed architecture isn't just inconvenient when working with AI — it's prohibitively expensive.

There is no one-size-fits-all architecture, however. What works in one context will work against you in another. You'll learn a wide range of patterns in this class, and how to combine them to create an optimal architecture for the problem at hand.

You'll also learn to design agentic systems – architectures where AI agents do the work — which are rapidly becoming a core part of modern software.

Outline

1/5 Architecture Fundamentals

The role of the Architect
Architectural and systems thinking
Incremental architecture
Integrating coding and product development
The Inspect & Adapt loop
Architectural patterns
Working with architectural trade-offs
Defining -ilities
Architectural Decision Records (ADRs)

2/5 Structural Architecture

Conway's Law
Layered architectures (just say no)
Hexagonal architecture
Atomic architecture
Distributed architecture
Deutch's Fallacies of Distributed Computing
Inter-service Communication
Idempotency
Logging, monitoring, and observability
Cloud-native architectures (containerization, lambdas, data in the cloud)

3/5 Story-Focused Architecture

User Stories and Problem Statements
The 4 C's: Communication, Collaboration, Confirmation, and Cards
Vertical Slice Architecture
API design for orchestrated systems
Deriving APIs from stories
Converting stories to a Domain Model
Component architecture and Microservices
Opaque components, impermeable boundaries
Single responsibility
Liskov substitution
Data (and database) encapsulation
Fault tolerance: Circuit Breakers and Bulkheads
Microservice characteristics that differ from other components
Micro front-ends
Agent and Agentic architectures – Presentation, Abstraction, Control

4/5 Domain-Focused Architecture

Event-driven architecture
Choreographed vs Orchestrated systems
Brokers
Queues and Publishers
Event payloads
User, transaction, and request IDs
Lightweight messaging systems
ZeroMQ (brokerless)
Kafka (brokered)
REST-to-messaging portals
Transaction management
Error detection and recovery
Self-scaling systems
Naming services
Domain-Driven Design
Event Storming

5/5 AI–Native Architecture

Micro-Agents and Macro-Agents
LLM integration and communication
Orchestration and the Judge
Talking to LLMs through public interfaces
The Model Context Protocol
Local vs Cloud LLMs
Selecting a model
Ollama, Pydantic
Risks and guardrails
Data vulnerabilities
Prompt injection
Context engineering and RAG
... and much more

By the end of this workshop, you'll master the effective software architecture principles, practices, and patterns that work well in the age of AI. You'll also get comfortable with AI-assisted software engineering and learn to integrate LLM-generated code into your applications efficiently while managing context, costs, and risks.

FAQ

For whom is this course?

This course is ideal for developers, architects, and tech leads with 3+ years of experience in enterprise software development.

How is the learning process organized?

The course is a mix of slides, demos, discussions, and exercises. Expect a safe environment with plenty of interaction, where all questions and ideas are welcome.

Do I need to prepare for this course?

You'll need to install a number of tools and libraries on your laptop. We'll send detailed instructions a week ahead of time — just make sure you have root access on your machine before then.