What it does This repository holds five Jupyter-based courses that walk developers through the Claude SDK and prompt engineering. The sequence runs from API basics (keys, streaming, multimodal prompts) through prompting techniques, real-world application, evaluation methods, and finally tool use. AWS and Google Vertex variants exist for some courses.

The interesting bit The courses deliberately target Claude 3 Haiku, Anthropic’s cheapest model, to keep student API bills minimal. It’s a rare case of a vendor optimizing its own educational materials for cost rather than upselling to its flagship model.

Key highlights

• Five sequential courses: API fundamentals → prompting → real-world application → evaluation → tool use
• Jupyter Notebook format: run the examples directly
• AWS Workshop and Google Vertex forks for cloud-specific learners
• Explicitly designed around low-cost Haiku to reduce follow-along expenses
• Covers production concerns like prompt evaluation, not just toy examples

Caveats

• No topics tagged on the repo, which makes discovery harder than it should be
• README is a flat list; course depth and prerequisites are unclear without clicking through

Verdict Worth bookmarking if you’re onboarding a team to Claude or need structured prompt-engineering practice. Skip if you already ship production LLM features and have your own evaluation frameworks.