AI Free course · Certificate included
Advanced Prompt Engineering
Context engineering, structured outputs, thinking controls, tool and agent prompts, caching, evals and optimisation for production
- Advanced
- 6 h 23 min
- 17 lessons in 7 modules
- 2 h 11 min of video lectures
- Updated Sep 2026
About this course
This advanced course turns prompting into an engineering discipline for teams shipping AI features in 2026. You will design system prompts and practise context engineering: deciding exactly what the model sees, in what order, and what to leave out. You will build few-shot sets and schema-enforced structured outputs, decompose work into chains, and learn when reasoning models and extended or adaptive thinking pay off, with hands-on effort controls in the Claude, OpenAI and Gemini APIs. A dedicated module covers tool definitions and system prompts for agents, including long-running context management. You will reduce hallucinations, defend against prompt injection, build evaluation sets, calibrate LLM judges and use automated prompt optimisation safely. Finally, you will run prompts like production code: versioning, prompt caching, cost and latency. Every lesson includes copy-paste Python you can run today.
Tools you’ll use
- Claude API
- OpenAI API
- Gemini API
- Python
- Pydantic
- DSPy
- Promptfoo
Skills
- Context engineering
- Structured outputs
- Reasoning models
- Agent and tool prompting
- LLM evaluation
- Prompt injection defence
- Prompt caching
- Prompt operations
What you’ll be able to do
- Design system prompts and curate context for accuracy, cost and latency
- Build few-shot sets and schema-enforced structured outputs with validation in code
- Choose when to use reasoning models and tune thinking effort with evidence
- Write tool definitions and agent system prompts that behave safely over long runs
- Reduce hallucinations and defend against direct and indirect prompt injection
- Build eval sets, calibrate LLM judges and run automated prompt optimisation safely
- Operate prompts in production with versioning, prompt caching and cost controls
Curriculum
Syllabus
- Modules
- 7
- Lessons
- 17
- Reading time
- 3.5 h
- Assessment questions
- 25
Design system prompts that behave like a clear job brief, then decide exactly what information the model sees, in what order, and why.
- Designing system prompts and rolesVideo lecture, 9′Video · 12 min
- Context engineering: what the model sees, and in what orderVideo lecture, 8′13 min
Use examples to teach patterns the instructions cannot, and get machine-readable outputs that downstream systems can trust.
- Few-shot and example designVideo lecture, 8′12 min
- Structured outputs and JSON schemasVideo lecture, 8′12 min
Break complex work into reliable steps, understand when explicit reasoning helps or hurts with modern reasoning models, and add verification loops.
- Decomposition and prompt chainingVideo lecture, 8′12 min
- Reasoning models: when step-by-step helps and when it hurtsVideo lecture, 7′13 min
- Self-critique and verification loopsVideo lecture, 8′11 min
Write tool definitions models use correctly, and system prompts that keep agents effective, safe and on-task over long runs.
- Tool definitions and tool-use promptsVideo lecture, 8′14 min
- System prompts for agents and long-running tasksVideo lecture, 8′14 min
Make models say only what they can support, admit uncertainty, and resist malicious instructions hidden in the content they process.
- Reducing hallucinations: grounding, citations and 'I don't know'Video lecture, 8′13 min
- Prompt injection and defence basicsVideo lecture, 8′13 min
Replace 'it looks good' with evidence: build evaluation sets, write rubrics, and use LLM-as-judge carefully.
- Building evaluation sets that reflect realityVideo lecture, 7′12 min
- Grading outputs: code checks, rubrics and LLM-as-judgeVideo lecture, 7′13 min
- Automated prompt optimisation and auto-promptingVideo lecture, 7′13 min
Run prompts like production software: versioned, tested, monitored and tuned for cost and speed.
- Versioning and managing prompts in productionVideo lecture, 7′11 min
- Cost and latency trade-offsVideo lecture, 7′12 min
- Prompt caching in practiceVideo lecture, 8′12 min
- Final assessment
Your certificate
Finish with a credential anyone can check
Earn the Certified Advanced Prompt Engineer badge: The holder can design production-grade prompts: precise system prompts, engineered context, few-shot examples and JSON schemas, prompt chains and verification loops. They can reduce hallucinations, defend against prompt injection, build evaluation sets with calibrated graders, and manage prompts with versioning and cost-latency trade-offs.
Completed all lessons and scored at least 80% on the final assessment.
- A public verification page
- A PDF certificate to download
- An Open Badge you can share
- One click to your LinkedIn profile
Final assessment
- 25questions drawn from a larger pool
- 40 mintime limit
- 80%pass mark
- 3attempts per 24 hours
Start learning today. It’s free.
Every lesson is free to read. A free account saves your progress, unlocks the final assessment and issues your certificate.