mircea

LLM Patterns (Working Paper)

A living catalogue of recurring patterns I keep running into when building software with and around large language models — the small, reusable shapes of solution that show up again and again across agents, tools, prompts, and evaluation.

It’s a working paper: a draft I’m developing in the open and revising as the ideas (and the tools) mature. Feedback is very welcome.

Status: working draft, June 2026

Read the working paper (Google Doc) →