Built an AI Wiki. Here's Why.

Got tired of explaining the same stuff to every client. So I built a wiki.

It's basically a collection of patterns I use when building production AI systems. RAG architectures, prompt engineering, model selection, deployment strategies. The boring stuff that actually matters.

The Problem

Every project starts the same way. Client wants to "use AI" but doesn't know what that means. You spend the first few weeks explaining basics instead of building things.

Docs are scattered everywhere. AWS has theirs, OpenAI has theirs, nobody agrees on anything. Half the tutorials are outdated.

The Solution

One place with everything I actually use:

• Architecture patterns that work
• Cost estimates that are realistic
• Deployment scripts I've tested
• Things that broke and how to fix them

What's In It

Mostly practical stuff. How to set up Bedrock properly. When to use RAG vs fine-tuning. How to not blow your budget on embeddings.

No fluff. No "AI will change everything" think pieces. Just code and configs.

Why Open Source

Because hoarding knowledge is dumb. Someone will figure it out anyway. Might as well save them time.

Plus it forces me to write things down properly instead of keeping everything in my head.

Check it out if you're building AI stuff. Or don't. I'm not your dad.