LLM-Assisted Domain Speedrun Learning

Overview

A structured protocol for using an LLM as a private conceptual tutor to rapidly bridge fundamental knowledge gaps in unfamiliar domains. The core principle: the LLM teaches concepts through metaphors and hints, never code — you write all code yourself in your own style.

This technique was battle-tested in a one-week Google interview preparation speedrun, taking someone from "unable to solve the simplest LeetCode problem" to completing 34 problems (18 Medium, 1 Hard) and advancing to on-site interviews.

When to Use

• You have a fundamental knowledge gap in a domain (not just rustiness)
• Previous learning attempts (books, videos, courses) have failed
• You're under a hard deadline (days to weeks, not months)
• You have a day job and limited study time
• The domain has learnable patterns (algorithms, system design, new programming paradigm, etc.)

When NOT to use: When you have time for a proper course, when the domain requires hands-on lab work an LLM can't simulate, or when the gap is purely about memorizing facts (use flashcards instead).

The Protocol