https://datatweets.com/courses/generative-ai/prompt-engineering/Recent content in Prompt Engineering on DATATWEETSHugoenCopyright (c) 2025 DatatweetsSat, 27 Jun 2026 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-1-anatomy-of-a-strong-prompt/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-1-anatomy-of-a-strong-prompt/A strong prompt is built, not wished. Dissect the anatomy of an effective prompt and see, on real model output, how each element changes the answer.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-2-sharpening-techniques/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-2-sharpening-techniques/A decent prompt works; a sharp prompt works every time. Learn the six techniques that make output predictable, each demonstrated on real model output in Python.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-3-few-shot-prompting-and-roles/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-3-few-shot-prompting-and-roles/Sometimes you can’t describe the output you want — you have to demonstrate it. Learn to teach the model by example with few-shot prompting, and to change its judgment with a role.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-4-structured-outputs/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-4-structured-outputs/Free-form ‘respond in JSON’ is fragile — stray prose, markdown fences, and invented field names break your parser. Learn two reliable ways to get structured output you can hand straight to a program.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-5-prompting-for-data-tasks/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-5-prompting-for-data-tasks/Prompting isn’t just for chat. Turn a language model into a data tool: extract structured fields from messy text, classify into a controlled vocabulary, summarize under constraints, and loop it over many items reliably.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-6-evaluating-and-improving-prompts/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-6-evaluating-and-improving-prompts/A prompt that ’looks good’ on one example often breaks on the next. Learn to measure prompt quality with a small test set, an automated check, an accuracy score, and an LLM-as-judge — then iterate on real numbers.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-7-reducing-hallucinations-and-unsafe-output/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-7-reducing-hallucinations-and-unsafe-output/A fluent answer is not a true one. Learn to ground a model in your own context, make it say ‘I don’t know’, ask it for quotes, and keep untrusted text from hijacking your instructions — all on real model output.https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-8-guided-project-prompt-toolkit/Fri, 14 Nov 2025 09:00:00 +0200https://datatweets.com/courses/generative-ai/prompt-engineering/lesson-8-guided-project-prompt-toolkit/The module capstone: build prompt_toolkit.py — a PromptTemplate plus ready-made summarize, classify, and extract tools, wired to a fixed model and low temperature, with a tiny eval harness that scores classify against labeled examples.