Prompt Engineering

A short, practical track for writing prompts that hold up across models and edge cases.

1. Step 01

   How models read

   Tokens, context, instruction hierarchy and why phrasing changes output.

   LLMsTokens

2. Step 02

   Core techniques

   Role framing, explicit constraints, few-shot examples and step decomposition.

   Few-shotCoT

3. Step 03

   Structured output

   JSON schemas, function calling, validation and repair loops.

   JSONSchemas

4. Step 04

   Creative prompting

   Image, video and design prompts: subject, style, lighting, camera and negative terms.

   MidjourneyVeo

5. Step 05

   Reliability

   Test sets, regression checks, guardrails and handling refusals or hallucination.

   Evals

6. Step 06

   Systems

   Prompt libraries, versioning, templating and cost/latency trade-offs in production.

   Templating