Learning

A self-contained course on the harness engineering that gives LLMs state, action, and limits, dissected across 24 sections and three real systems.

Official code for a book that ships PicoAgents, a full multi-agent framework written from scratch so you can see how the pieces fit together without vendor abstraction getting in the way.

A curated attempt to keep the sprawling DeepSeek Harness plugin ecosystem navigable.

DeepSeek's agent runtime has architecture docs; this is the missing field manual for actually using it.

A Chinese-language full-stack course that treats copy-paste coding as a failure mode, making readers hand-craft `Llama 2` and `Transformer` blocks before moving on to fine-tuning and deployment.

Curated codes and references for a Portuguese-language postgraduate program in applied AI, organized from browser machine learning to local LLMs and MCP automation.

These prompts exist because treating ChatGPT like a magic template machine is why your user stories still sound like Mad Libs.

A game developer's personal learning journal, open-sourced: books, videos, tools, and models that actually helped someone build with AI.

A curated collection of Jupyter notebooks and R Markdown files covering the standard ML curriculum, useful mostly as a reference for how to structure your own.

A curated anthology of the most effective—and verbose—GPT Image 2 prompts scraped from X.

Someone had to catalog the explosion of tools trying to herd AI coding agents in parallel.

A hands-on Node.js tutorial series that makes you implement embeddings, vector stores, and retrieval yourself so RAG stops feeling like magic.

A project-based course that teaches LangChain v1 and LangGraph by building real agents with real APIs.

Because launch posts describe the vendor’s architecture, not your machine’s permission prompts and credit-card charges.

Because choosing between nineteen agent frameworks shouldn't require nineteen browser tabs.

A single, durable archive for the 2018 CS229 lectures, notes, and problem sets before they scatter across the internet.

A systematic, open-source knowledge base that closes the gap between scattered CUDA tutorials and actual AI infrastructure engineering.

Not prompt tricks—control structures: how Claude Code and Codex constrain, recover, and govern model behavior inside real workflows.

It catalogs system prompts, jailbreak techniques, and sandbox reconnaissance to teach researchers how LLMs are actually instructed.

It sorts the sprawling open-source landscape of AI companions into a tagged index of chat clients, memory engines, and virtual phones.