AI & Prompt Engineering Roadmap

Build the right mental model for how AI tools actually work before using them.

• What large language models are and how they're trained
• Tokens, context windows, and why they matter
• The difference between GPT-4, Claude, Gemini, Llama
• Hallucinations — why they happen and how to spot them
• AI capabilities and current limitations

Learn to communicate clearly with AI to get consistently better outputs.

• The anatomy of a good prompt — context, instruction, output format
• Role prompting and system instructions
• Chain-of-thought and step-by-step reasoning
• Few-shot examples — teaching the model with context
• Iterative prompting — refining outputs
• Temperature and other generation settings

Use AI tools to do your existing work 3–5x faster.

• AI writing — research, drafting, editing, summarising
• AI for email — writing and reply suggestions
• AI for presentations — Gamma, Beautiful.ai
• AI for research — Perplexity, NotebookLM
• Meeting transcription and summarisation — Otter.ai
• Building custom GPTs and AI assistants

AI-assisted coding is a superpower. Learn to use it effectively.

• GitHub Copilot — autocomplete and chat
• Cursor — chatting with your codebase
• Claude Code — agentic terminal-based coding
• AI code review and refactoring
• Generating tests with AI
• Bolt.new and Lovable for no-code app building

Go from using AI to building with it. Create your own AI-powered features and applications.

• OpenAI and Anthropic API setup
• Chat completions — messages, roles, and parameters
• Streaming responses for better UX
• Tool use and function calling
• RAG — Retrieval-Augmented Generation basics
• Prompt caching for cost reduction
• Deploying an AI feature to production