How it is organised

The map comes first: one enterprise system drawn end to end, where every component carries what has been written about it. The journal holds everything written, newest first. Series are short runs to read in order, and paths pick a route through them from a problem to production. Everything is also read through nine engineering disciplines:

  • AI & Models · What does the model actually do, and where does it need the system to help?
  • Data & Databases · Where does data live, how does it move and how fresh is it?
  • Networking · Which paths does traffic take, and where does it leave your network?
  • Security · Where are the trust boundaries, and what can cross them?
  • Identity & Access · Who is acting, on whose behalf, with which permissions?
  • Integration · How does this connect to the systems the business already runs?
  • Compute & Infra · What runs where, and how does it scale and fail?
  • Observability & Ops · How do you know it is working, and what do you do when it is not?
  • Cost & Governance · What does it cost, who approved it, and what is on the record?

The spine

Everything hangs off a map. Each component on a map carries the entries, code, tools and references that belong to it, and paths follow one topic from problem through architecture, decision, code and production. Reading one thing should always show you the next.

What you will not find

Winner scores, industry rankings or polished success stories without the failure that preceded them. The radar is my experience in specific systems. The failure stories come from real work, with names, numbers and details changed so no system or employer can be identified.

On the bench

What I am working on, updated Oct 9, 2026.

  • ShippingUnseen UI 0.1.0-alpha.3. Both packages are on npm, and this site runs on them. Next up: router-aware link components.
  • ResearchingMCP gateways. Permissions, identity and audit between agents and tools.
  • DocumentingEnterprise RAG. Retrieval that stays honest after the demo.

Who writes it

Vamsi Krishna, an engineer building AI systems and the tooling around them. Opinions are earned in production and revised when the evidence changes.

Get in touch

Open an issue or discussion on GitHub. To follow new entries, subscribe by email (one email per new map or story, no digest) or use the RSS feed.