Become an AI engineer
For people who write code: how models work and how to build with them. 445 lessons in 52 units, from level 1 to level 5.
Units
- How models work, plainly
- The transformer, plainly
- Tokenization and data, plainly
- Scaling, plainly
- Post training, plainly
- Reasoning, plainly
- Truthfulness, plainly
- Retrieval and tools, plainly
- Safety, plainly
- Foundations
- The transformer
- Tokenization and data
- Scaling
- Post training
- Reasoning
- Truthfulness
- Retrieval and tools
- Interpretability
- Safety
- Working with Claude
- Programmatic Tool Calling: letting code call your tools
- Working with ChatGPT
- Realtime API: streaming a live voice conversation
- Agents and harnesses
- Claude Code
- Coding agents
- Agent Metadata API: reading what the agent knows, right now
- GitHub
- Does it work?
- Perplexity: how surprised is the model?
- Cost and limits
- Agentic harness
- Inference engineering
- Training infrastructure
- Model optimization
- Data engineering
- LLM security
- LLMOps
- GPU programming
- Open models
- How tools connect
- Claude (platform)
- Claude Managed Agents: define an agent, run a session, collect its files
- Images, audio and video
- Contrastive learning: pick the true match out of the crowd
- Cloudflare
- Claude connectors
- UI State Management: what the model sees, what the UI keeps, what the server stores
- Copilot
- ServiceNow Identity Sync: the user_changes resource
- ElevenLabs
- Custom LLM Integration: Bringing Your Own Model
- Groq
- Getting started with Groq
- LangChain
- Getting started with LangChain
- Modal
- n8n
- Sending bytes as bytes: binary payloads, PCM audio and image blocks
- Replicate
- Getting started with Replicate
- ComfyUI: building image workflows from nodes
- Runway
- Runway Avatar API: adding a talking Avatar to your app
- Supabase
- Vercel