Prompt Engineering Bootcamp
Get dramatically better outputs from LLMs
About this course
Master the craft of talking to large language models. Frameworks, patterns, and pro tricks that turn vague prompts into reliable results.
What you'll learn
- Apply the CRISP prompt framework
- Use few-shot, chain-of-thought, and role prompting
- Debug bad outputs systematically
- Build reusable prompt templates
Curriculum
Build a Prompt Library for Your Job
Pick 5 recurring tasks at your job (e.g. write a status update, summarise a Zoom transcript, draft a follow-up email, generate SQL from a question, classify a support ticket). For each, ship a versioned, tested prompt template: 1. **Goal** in plain English (1 sentence). 2. **Template** with named variables: `{{customer_name}}`, `{{order_id}}`. 3. **3 example inputs** and the expected output for each. 4. **One known failure mode** with a mitigation (e.g. 'model invents prices → add `Use ONLY data from the table below'`). 5. **Token estimate** per call and rough monthly cost at 100 calls/day on Claude Sonnet 4.5. Submit as a single markdown doc.
- 15 distinct, real-world tasks (not toy examples)
- 2Templates use named variables, not free text
- 3At least one failure mode + mitigation per prompt
- 4Token + cost math present and reasonable
- 5Library is reusable by a teammate who didn't write it
What graduates shipped.
Once you ship, your project lives here — linked from your public verify page so recruiters can click through.