AI Tools & Concepts
Complete guide for software engineers — from how LLMs work to building with AI APIs, agents, and RAG
Part 1 — Core Concepts
What large language models actually are, how they think
What tokens are, why they matter, how to count them, cost
What it is, why it's limited, how to manage it, pricing
How AI generates text, randomness controls, when to tune
How AI understands meaning, similarity search, RAG basics
Three ways to customize AI, when to use which
Part 2 — Prompting & Getting the Best Output
Zero-shot, few-shot, chain-of-thought, system prompts
ReAct, tree-of-thought, self-consistency, structured output
How to write effective system prompts, persona, constraints
Part 3 — AI Tools Ecosystem
Models, features, API, Claude Code, artifacts, projects
GPT-4o, o1/o3, API, plugins, custom GPTs
GitHub Copilot, Cursor, Windsurf, Cline, Claude Code
Llama, Mistral, Gemma, DeepSeek, running locally
Part 4 — Building With AI
Anthropic API, OpenAI API, structured output, streaming
How AI calls tools, MCP protocol, agentic patterns
What agents are, agent loops, multi-agent systems, frameworks
Vector DBs, chunking, retrieval, end-to-end RAG pipeline
How to measure AI quality, evals, benchmarks, red-teaming
Part 5 — Staying Current
Who's who, model comparison, pricing, capabilities
Latest AI announcements and developments
Notes & Resources
Key Terms Quick Reference
| Term | Definition |
|---|---|
| LLM | Large Language Model — neural network trained on text to generate text |
| Token | Smallest unit of text the model processes (~4 chars or ~3/4 of a word) |
| Context Window | Maximum tokens the model can see at once (input + output) |
| Temperature | Controls randomness: 0 = deterministic, 1 = creative |
| Embedding | Vector representation of text that captures meaning |
| RAG | Retrieve relevant docs, feed them to LLM for grounded answers |
| Agent | AI that can use tools, make decisions, and take actions in a loop |
| MCP | Model Context Protocol — standard for connecting AI to tools/data |