asela38

# LLM Wiki at Work

A pattern for building a **work-context second brain** using LLMs.

This is an idea file; it is designed to be copied and pasted into your own LLM Agent (Claude Code or similar) to bootstrap a new vault dedicated to office/project/professional work. Its goal is to communicate the high-level pattern; your agent will build out the specifics in collaboration with you. It is a sibling of the personal-knowledge version described in [[llm-wiki]].

## The core idea

At work, you read and generate enormous amounts of textual material every day: meeting notes, standup updates, design documents, code reviews, tickets, Slack threads, architectural decisions, stakeholder emails, and AI-assistant conversations about your own codebase. Most of it is written to be forgotten. When the same question comes up three sprints later, you re-derive the answer — or you ask a colleague who re-derives it — or you stare at the pull request that resolved it and can't remember why the resolution made sense. Professio

karpathy

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

timotheehub

TransferWise Coding Challenge #1

Editorial of the first question: TransferWise Invite Program Winner

The questions are still open and you can try them here.

The question was about finding the person who invited the largest number of users through TransferWise's invite program. This question can be generalized to finding the largest tree in an acyclic unidirectional disconnected graph. If you don't remember it, you can also find it below.

There are multiple ways to solve the first question including:

1. Finding the root nodes of the trees by finding the nodes without parents

ms-tg

/**
 * ```
 * Does JDK8's Optional class satisfy the Monad laws?
 * =================================================
 *   1.  Left identity:  true
 *   2.  Right identity: true
 *   3.  Associativity:  true
 *
 * Yes, it does. 
 * ```