convez.ai
MCP ServerProdutividadeIniciante
conorbronsdon

AI Learning Resources

conorbronsdon/ai-learning-resources

Caminho estruturado de zero a builder de AI, sem papo de vendedor

Ver no GitHub
InicialPluginScore · de 0 a 100

Repositório com caminho de aprendizado estruturado em 4 estágios, desde conceitos fundamentais de IA até construção de workflows e agents com Claude Code e MCP. Prioriza qualidade sobre quantidade, combinando vídeos, artigos e cursos estabelecidos.

Instalação

$npx convez-ai conorbronsdon/ai-learning-resources

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
2
Forks
1
Acompanhando
0
Saúde
42%
Último commit
há 1 dia
Versão
—
Issues abertas
3
Contribuidores
1

Casos de uso

  • Aprender como funcionam LLMs e modelos de IA
  • Transição de usuário para desenvolvedor de ferramentas com IA
  • Entender prompteng, finetuning e construção de agents
  • Estudar arquitetura de transformers e redes neurais

Veredito

Recurso sólido e bem organizado para iniciantes que querem entender IA em profundidade antes de construir. Curadoria de qualidade com links verificados, mas é um repositório de referências — funciona melhor como guia de estudo do que como ferramenta ativa de integração.

README

A tradução para o português deste README ainda está na fila.

AI Learning Resources

A curated learning path from zero to building your own AI-powered workflows, apps, and agents.

GitHub stars License: MIT X


This isn't a dump of links. It's a path. Start at Stage 1 and work forward, or jump to wherever you are.

Last updated: June 2026

Contents

Prerequisites

Git: Claude Code and many AI tools use Git for version control. If you're new to it, GitHub's About Git guide will get you up to speed in 10 minutes. Prefer video? Git Explained in 100 Seconds (Fireship, 3 min) covers the essentials.

Prefer a structured course?

If you'd rather follow a full curriculum than pick through curated links, start with one of these:

CourseProviderFormatLevel
Anthropic AcademyAnthropicSelf-paced, freeBeginner → Advanced
OpenAI AcademyOpenAISelf-paced, freeBeginner
Generative AI for BeginnersMicrosoft21 lessons, Python/TSBeginner → Intermediate
AI for BeginnersMicrosoft12-week curriculumBeginner
LLM CourseMaxime LabonneColab notebooksIntermediate → Advanced
Start Machine LearningLouis BouchardReading listBeginner

Otherwise, follow the staged path below.


Stage 1: Understand How AI Works

You don't need a CS degree. You need mental models for what's happening under the hood so you can reason about what AI can and can't do.

Start here

  • Deep Dive into LLMs like ChatGPT — Andrej Karpathy (3.5 hrs) The single best general-audience walkthrough of how LLMs are built: pretraining, finetuning, RLHF. No technical background needed. If you watch one thing, make it this. Shorter version: Intro to Large Language Models (1 hr)

  • Neural Networks — 3Blue1Brown (video series) The gold-standard visual explainer. Starts with "what is a neural network?" and builds through backpropagation to transformers and attention. Chapters 5-7 cover LLMs specifically.

  • What Is ChatGPT Doing... and Why Does It Work? — Stephen Wolfram Deep, readable explainer covering embeddings, tokens, transformers, and the philosophical "why" behind next-token prediction. Also a book.

Go deeper

  • The Illustrated Transformer — Jay Alammar The canonical visual blog post for understanding transformer architecture. More technical than the above, but still accessible with diagrams for every step.

  • The AI Revolution: Road to Superintelligence — Tim Urban (Wait But Why) Written in 2015, pre-LLM, but still the best accessible piece on why AI matters at civilizational scale. Good for the "why should I care" framing.

  • GenAI Handbook — William Brown Curated roadmap organizing the best scattered resources into 9 structured sections. Good if you want a self-directed syllabus.

Podcast episodes

Episodes from Chain of Thought that help build your mental model:

  • Beyond Transformers — Maxime Labonne. Model architectures beyond the transformer — what comes next.

Stage 2: Start Using AI Effectively

You understand the basics. Now make it useful.

Choosing a tool

There are many AI tools, and more launching every week. The main ones worth knowing:

  • ChatGPT (OpenAI) — the most widely used, good general-purpose starting point
  • Claude (Anthropic) — strong at writing, analysis, and long documents (my personal favorite as of 2026)
  • Gemini (Google) — integrated with Google Workspace
  • Perplexity — AI-powered search with citations
  • Claude Code / Codex / Cursor / Windsurf — AI coding tools (see Stage 3)

Don't overthink the choice. Pick one and start using it daily. You'll develop preferences fast.

Start here

  • The Practical Guide to Using AI to Do Stuff — Ethan Mollick The canonical "OK I have ChatGPT, now what?" post. Covers writing, analysis, images, code, and when AI fails. Regularly updated. Video alternative: AI's Productivity Paradox — and What It Means for You (TED Talk, 12 min) — live demos of AI transforming productivity.

  • How I use LLMs — Andrej Karpathy (2 hrs) A practical tour of his day-to-day LLM use: which tool for which job, settings worth changing, voice, search, and file workflows. The hands-on companion to his Stage 1 deep dive.

  • Co-Intelligence: Living and Working with AI — Ethan Mollick (book) Best single book for non-technical professionals. Four principles: always invite AI, be the human in the loop, treat it as a smart-but-alien collaborator, assume it's the worst AI you'll ever use.

Go deeper

Know the limits

AI tools hallucinate, leak data to providers, and confidently produce wrong answers as they seek to predict the right response. Before relying on them:

  • Never paste sensitive/proprietary data into a tool without understanding its data policy
  • Always verify factual claims, especially numbers, citations, and dates
  • AI is a collaborator, not an authority — you own the output

Podcast episodes

Stage 3: Build with AI Coding Tools

Hands-on. Install tools, build workflows, connect AI to your actual work. This section focuses on Claude Code and MCP, but the patterns apply to Cursor, Windsurf, Cline, and other AI coding tools too.

Claude Code

  • Claude Code Overview — Anthropic (official docs) Start here. Installation (terminal, VS Code, JetBrains, desktop, web), core capabilities, and links to everything else. Video alternative: Claude Code Beginner's Tutorial (Peter Yang, ~15 min) — hands-on walkthrough from install to building a working app.

  • Claude Code Quickstart — Anthropic Walk through your first task: explore a codebase, make changes, commit.

  • Claude Code Best Practices — Anthropic Patterns for getting the most from Claude Code — CLAUDE.md setup, workflow patterns, sub-agents.

  • Claude Code in Action — Anthropic Academy Free official course with certificate. Teaches building, configuring, and sharing Skills.

  • Anthropic Academy — Anthropic 13 free self-paced courses: Claude 101, AI Fluency, Claude Code, MCP, the API, and certification (Claude Certified Architect).

  • How Anthropic Teams Use Claude Code — Anthropic (PDF) Internal practices showing how Anthropic's own teams use Claude Code. Good for advanced patterns.

  • Claude Code GitHub repo — Anthropic Open source. Good for architecture understanding, CLAUDE.md examples, and filing issues.

  • Claude Agent SDK — Anthropic Build your own agents on the same tools and agent loop that power Claude Code, in Python or TypeScript. The natural next step once skills and MCP servers feel comfortable.

  • Building Effective Agents — Anthropic The reference piece on agent design: when a simple workflow beats an agent, and the composable patterns (prompt chaining, routing, orchestration) that production systems actually use. Read this before building anything autonomous.

  • MARVIN template — Sterling Chin Reference architecture for an autonomous background agent that runs on GitHub Actions + Claude Code. Good example of what's possible once you're comfortable with the basics.

MCP (Model Context Protocol)

  • What is MCP? — Anthropic / MCP team "USB-C port for AI." What MCP enables, why it matters, and who supports it (Claude, ChatGPT, VS Code, Cursor, Gemini). Video alternative: Model Context Protocol Explained (Fireship, ~8 min) — fast-paced explainer of what MCP is and why it matters.

  • MCP Quickstart: Build a Server — MCP team Official tutorial: build a weather MCP server and connect it. Python and TypeScript.

  • MCP Architecture & Concepts — MCP team Core concepts: resources, tools, prompts, transports, client-server architecture.

  • Official MCP Servers — MCP team Reference implementations and community servers. Browse to see what MCP servers look like in practice.

  • awesome-mcp-servers — wong2 Largest curated list of MCP servers by category. Good for discovering what's available.

What to build first

Once you have Claude Code installed, try these as starter projects. Don't have a project? Create an empty folder with a README — that's enough to start.

  1. Add the avoid-ai-writing skill to a project and run it on something you've written. Seeing Claude audit your own text is the fastest way to understand how skills work.
  2. Write a CLAUDE.md file for a project you're working on. Start with 3-5 instructions about how you want Claude to behave. Iterate from there.
  3. Install an MCP server (start with one from the official servers repo) and use it in Claude Code. The jump from "Claude can read files" to "Claude can access my calendar" is when it clicks.

Skills & Tools

  • agentskills.io — Portable skill format that works across 40+ AI coding tools.
  • AI Tools for Creators — Curated collection of production-tested skills and MCP servers for content creators and knowledge workers.

Podcast episodes

Stage 4: Think Strategically About AI

The hardest part isn't using AI — it's knowing what it changes about your work, your industry, and your decisions.

Start here

Podcast episodes

Keep learning

Ongoing sources worth following once you've worked through the stages:

  • One Useful Thing — Ethan Mollick's Substack. The best ongoing writing about AI's practical impact on work. Updated regularly.
  • Chain of Thought newsletter — Weekly updates on AI tools, workflows, and what's working in production.
  • How I AI — Podcast focused on how people actually use AI in their daily work. Practical, not theoretical.
  • Dwarkesh Podcast — Long-form interviews with AI researchers, founders, and thinkers. The best deep-dive conversations on where AI is heading.
  • Simon Willison's Weblog — Deep, practical coverage of LLMs, tools, and the developer experience side of AI.

About

Maintained by Conor Bronsdon. I host Chain of Thought, a podcast about AI infrastructure, developer tools, and how practitioners use AI in production. Many of these resources were discovered through conversations with guests on the show.

Subscribe to the Chain of Thought newsletter for weekly updates on AI tools, workflows, and what's actually working in production.

Contributing

PRs welcome. To suggest a resource, open an issue or submit a PR — include a sentence on why it's valuable and where it fits in the learning path. Quality bar: you've used it and it actually helped.


Disclaimer

All views, opinions, and statements expressed on this account are solely my own and are made in my personal capacity. They do not reflect, and should not be construed as reflecting, the views, positions, or policies of Modular. This account is not affiliated with, authorized by, or endorsed by Modular in any way.

License

MIT (see LICENSE).

Ver o original no GitHub

Bom para

Relacionados