Engineering on Morteza Azizi

The First Architectural Decision Wasn't About AI

Wed, 26 Aug 2026 00:00:00 +0000

The First Architectural Decision Wasn’t About AI Link to heading

Part 3 of the Architecture Through Engineering series.

The First Decision Link to heading

The project was supposed to be about RAG.

Documents.
Chunks.
Embeddings.
Retrieval.
LLMs.

But before any of that, there was a more basic question:

What depends on what?

If the API knows about the database, the application knows about HTTP, and the domain knows about infrastructure, technical decisions start leaking across the system.

The Shape of the System

Sun, 09 Aug 2026 00:00:00 +0000

The Shape of the System Link to heading

Part 2 of the Architecture Through Engineering series.

The first two articles in this series were about why I chose RAG and why I wanted to build a reference implementation in the first place.

Now it’s time to build something.

Not the full system.

Not yet.

Before choosing a vector database, an embedding provider, or an LLM, I need to answer a much simpler question:

Why I Chose RAG as My First Reference Implementation

Mon, 03 Aug 2026 00:00:00 +0000

Why I Chose RAG as My First Reference Implementation Link to heading

Part 1 of the Architecture Through Engineering series.

The Obvious Question Link to heading

After publishing the manifesto, one question appeared almost immediately.

Why RAG?

It’s a fair question.

The internet already has thousands of RAG tutorials, examples, and reference implementations. If the goal of this series is architecture, I could have chosen almost anything.

Architecture Is Easy Until You Have to Build It

Tue, 28 Jul 2026 00:00:00 +0000

Architecture Is Easy Until You Have to Build It Link to heading

Part 0 of the Architecture Through Engineering series — the manifesto behind it.

Introduction Link to heading

There is no shortage of articles about software architecture.

Open any technical blog, attend a conference, or browse LinkedIn, and you’ll find countless diagrams, principles, frameworks, and opinions about what good architecture looks like. Most of them are written the same way: a clean model, a confident conclusion, and very little of the mess that came before it.