Skip to content

GenAI & Agentic AI — Exam 2 Prep

A beginner-friendly study guide for the 2nd proctored exam of the upGrad × IIT Kharagpur course on Generative and Agentic AI. It covers only the sessions taught after the 1st exam.

Exam format

24 multiple-choice questions in 90 minutes. Students from the first exam slot reported some formula-based calculation questions from the fine-tuning chapters. Every formula from the slides, with worked examples and calculation practice, is collected on the Numbers & formulas page.

How to use this guide

  1. Read each chapter's "In one minute" and "Exam radar" first. The radar marks topics 🔴 High, 🟠 Medium or 🟢 Low likelihood, based on how much class time they got, how often they appear across the pre-read, slides and post-read, and how many student questions they drew.
  2. Go through the concepts. Each one has a plain-language explanation, an everyday analogy, a small example and the common misunderstandings.
  3. Finish with the practice questions. Every chapter ends with 5–10 MCQs that give instant feedback and explain every option. Got one wrong? The explanation points you to the section to re-read.
  4. Then take the practice mocks. There are eight timed papers. Mocks 1–4 are the official practice set, split into four. Mocks 5–8 are original and each one is deliberately different, so you don't get used to a single pattern.

Chapters

Part A · Fine-tuning

Part B · Vision-language models

Part C · Agents

Part D · Responsible AI

Exam toolkit (use these in the final week)

  • Focus map: every high-likelihood topic, ranked, plus a rough question split across the four parts
  • Which one when?: "if the scenario says… → choose…" decision rules
  • Commonly confused pairs: look-alike concepts and how to tell them apart in an MCQ
  • Numbers & formulas: every formula from the slides, worked step by step, with 12 calculation questions
  • Cheat sheet: last-hour revision, chapter by chapter
  • Glossary: every term, defined in one line

About sources

Chapters are built from the live-class slides, pre-reads, post-reads, notebooks and Q&A sheets. Where the course material had gaps (for example, Responsible AI had no slides), we used public sources and marked those parts as beyond the course slides.