Prompt Engineering, the Full Walkthrough

We took the same lazy "write me an essay" prompt every ChatGPT beginner uses and rebuilt it into the kind of structured, reliable, production-grade prompts the engineers behind real AI products are actually shipping. Our instructor Dr Akshika (Aki) Wijesundara led the session, and by the end the model was returning clean JSON every time, controlling its own reasoning depth, calling external functions on cue, and handling edge cases without breaking. Then we took the detour everybody's curious about: how LLM jailbreaks actually work. 🎓 JOIN THE NEXT COHORT → https://theaiinternship.com/ • 12 weeks, live mentor-led sessions • Build real projects: RAG, agents, fine-tuning, MCP, capstone • Small batches, direct access to instructors • Career support + portfolio reviews ━━━━━━━━━━━━━━━━━━━━━ In this session you'll see: ✓ Why prompt engineering is still alive (and more important) in 2026 ✓ The full taxonomy: zero-shot, few-shot, chain-of-thought, role prompting ✓ Structured outputs — JSON schema, response_format, when to enforce ✓ Reasoning effort + verbosity controls (when to crank, when to dial back) ✓ The OpenAI API surface: Response API vs Chat Completions ✓ Function calling for plugging the model into real tools ✓ Common pitfalls — malformed JSON, verbose answers, ambiguous roles, hallucinations — and the fixes ✓ A short detour into how LLM jailbreaks actually work 🕐 CHAPTERS 0:00 Intro — today's focus is prompt engineering 1:42 Today's agenda 3:23 What prompt engineering actually is 5:37 Prompt techniques — overview 6:26 Few-shot prompting (with examples) 6:57 Chain-of-thought ("think step by step") 7:50 Role prompting (you-are-a-financial-analyst style) 8:21 Zero-shot prompting 10:15 Best practices — clarity + limiting ambiguity 10:53 System prompts 14:38 Testing your prompts iteratively 16:10 Enforcing output format (JSON only) 16:37 Delimiters and boundaries in prompts 17:36 Chain reasoning + breaking prompts into smaller pieces 22:00 Q&A — does context carry across conversation? 23:24 Conversation context management 25:03 Chain-of-thought — live walkthrough example 31:20 OpenAI models — O-series (reasoning) vs GPT-4.x 31:49 GPT-4.1 — smartest non-reasoning model 38:03 Live demo — prompting from inside Cursor 40:23 Response API vs Chat Completions 40:58 Reasoning effort 41:02 Verbosity controls 41:50 Structured output 44:45 JSON schema — the key takeaway 46:09 Common pitfalls + fixes (malformed JSON, over-verbose) 47:45 Hallucinations — and how to constrain them 48:31 Homework + next steps 49:40 Q&A — org-level system prompts, prompt management 51:40 Q&A — verbosity vs reasoning effort 55:33 Q&A — connecting external tools (MCP teaser) 56:00 Wrap 🛠️ TOOLS MENTIONED • OpenAI Response API — https://platform.openai.com/docs/api-... • Chat Completions API — https://platform.openai.com/docs/api-... • OpenAI Prompting Guide — https://platform.openai.com/docs/guid... • OpenAI Playground — https://platform.openai.com/playground • Cursor (used in demo) — https://cursor.com ━━━━━━━━━━━━━━━━━━━━━ 📚 THE FULL CURRICULUM W01 — Environment Setup & Your First OpenAI Call W02 — Build a ChatGPT Clone (Two Ways) W03 — REST APIs, JWT & FastAPI (Unhackable Backend) W04 — Vector Embeddings & Semantic Search W05 — Fine-Tuning ChatGPT (Turn It Into Your Company Intern) W07 — RAG: How to Stop Your AI from Making Things Up W08 — Build an AI Agent the Proper Way (No Gatekeeping) W09 — MCP: Learn It Now or Get Left Behind W11–W12 — Capstone project + showcase → Apply: https://theaiinternship.com/ About TAI — The AI Internship We train engineers to ship real AI products. 12-week mentor-led cohorts, real codebases, real deployment, real career outcomes. #PromptEngineering #ChatGPT #LLM #AIEngineering #OpenAI #StructuredOutput #JSONMode #FunctionCalling #PromptDesign #BuildWithAI #TheAIInternship