Skip to content

Start Here

Welcome. This is a working reference for making AI engineering legible — built by and for Solutions Engineers and Architects who have to stand between a CTO and an engineer and make sure they're solving the same problem.

📝 What This Is (and What It Isn't)

This is not another deep-RAG tutorial competing on tooling mastery. The internet has a glut of those. The scarce skill in 2026 is translation: scoping an AI POC that won't embarrass anyone, framing build-vs-buy in dollars, explaining why the demo hallucinated without losing the room. That's the SE/SA job, and almost nobody publishes it well.

The thesis, in one line: "I can't out-engineer a Staff ML Engineer. But I can stand in front of your exec team and your engineers in the same meeting and make sure they're solving the same problem — here's how."

The defense against this being shallow is non-negotiable: the labs still build the real thing. You run the RAG pipeline, write the eval, measure the latency. The translation layer is added on top of rigor, never substituted for it.

🔥 If You're in a Live AI Deal Right Now

Jump straight to what you need:

🧭 If You're Orienting for the First Time

Read the foundations on-ramp in order — it assumes no prior AI knowledge and gets you fluent fast:

  1. How LLMs Actually Work (no math) — the one idea everything hangs on, plus why models hallucinate.
  2. The Four-Layer Map — the mental model the whole site is organized around.
  3. AI Vocabulary for SAs — the glossary you wish you'd had.
  4. LangGraph in 10 Minutes — how agent workflows are wired.

Past the foundations, see Learning Paths for four role-based routes through the rest of the site — SE/SA entering AI, live deal, hands-on, and DevOps-to-AI crossover.

🗺️ How the Site Is Built

Everything here is organized under the four-layer map (L1 apps & agents → L2 LLMOps & infra → L3 MLOps & data → L4 architecture & governance) and split into content types that each do one job well:

TypeTeaches byStatus
Foundationsorienting✅ available now
Decision framesdeciding (build-vs-buy, cost, "when X vs Y")✅ available now
POC playbooksscoping & de-risking an engagement✅ available now
Talk tracksthe customer-safe "say it like this" card✅ available now
Visualsseeing — translation diagrams✅ available now
Labsdoing — hands-on, local-first, mostly $0✅ Labs 01–07 live
Lessonsthe concept + the decision behind it✅ apps-agents, architecture-governance, mlops-data live

The build is phased so the SE/SA spine — the differentiators — comes first. See product/BUILD-PLAN.md for the full plan and AGENTS.md for the project's source of truth.

📋 The Conventions That Keep It Consistent

  • Depth Standard — how a generic page becomes a worked, talk-track-equipped one. See DEPTH-STANDARD.md.
  • Canonical Cast — the reused tool examples (Ollama, Qdrant, Langfuse, LangGraph…) that make the site read authored, not assembled. See CANONICAL-CAST.md.
  • Contributing — git workflow and the dead-link CI gate. See CONTRIBUTING.md.

🔗 The Trilogy

This is the third repo in a set with one through-line — a Solutions Architect who makes hard technical domains understandable. solutions-playbook (how I run engagements) → devops-studio (the infra I can speak to) → ai-engineering-studio (making the newest, least-understood domain legible). Each stands alone; together they tell a single story.


This site is under active build. The foundations on-ramp, SE/SA spine, and all five labs are complete; Phase 4 (architecture & governance) is wrapping up and Phase 5 (capstone, observability, trilogy polish) is next.