Learnstead
A Korean-language collection of guides, tutorials and hands-on labs that teach running local LLMs, building retrieval-augmented AI over personal documents, using AI coding agents responsibly, and designing agent skills, MCP tools and context engineering.
Learnstead is organized as a set of topics a learner can pick from independently: coding with AI while keeping changes reversible through git and worktrees, understanding databases and their backup and permission models, running a language model locally and connecting it to a program, building retrieval-augmented generation over personal documents and diagnosing its failures, and teaching an agent repeated procedures through skills, MCP and context engineering. Each topic's guides explain concepts and decision criteria, its tutorials build a working result along one path, and its hands-on labs compare expected and actual outcomes on a fixed task.
As of the current material it offers 11 guides, 2 tutorials and 8 hands-on labs, with practical exercises covering things such as building a local RAG system with Ollama and Python, connecting a local Supabase and PostgreSQL database to a sample signup application, and reproducing and diagnosing common RAG retrieval and generation failures against a labeled evaluation set.
It is written mostly as Python-based labs and tutorials and is released under the Apache-2.0 license.