AIMentor — AI Expertise Learning Platform

Untimed. Answers are graded when you submit, with explanations and spaced-repetition scheduling.

Question 1easy
What is the core requirement of the capstone project?
Maximum scale and noveltyCompleteness across the lifecycle — a real input produces a real, evaluated output in a deployed system, with the business case written upUsing the largest possible modelThe most lines of code
Question 2easy
What are the two outputs of the capstone?
A training dataset and a modelA working system AND a case study that becomes your portfolioA Dockerfile and a promptA slide deck only
Question 3medium
How should you scope the capstone to make it finishable?
Cover as many features as possibleOne workflow / user / document set / channel — a complete, narrow slice with depth across the lifecycleBuild the whole platform before shipping anythingSkip deployment to save time
Question 4medium
Which phases make a capstone a 'system' rather than a 'demo'?
Only the build phaseEvaluation/hardening (evals, guardrails) and deployment — the unglamorous tracks (5-7)A bigger promptMore tools
Question 5hard
In the build phase, if the task is a fixed retrieve-then-generate flow, what is the right Track-4 decision?
Always build a multi-agent systemUse a chain, not an agent — and be able to justify that call (known fixed steps)Use an agent to look advancedFine-tune a model instead
Question 6medium
Why is 'it runs in my notebook' insufficient for the capstone?
Notebooks are slowThe capstone proves you can ship — it must run in a real deployed environment, not just locallyNotebooks can't call APIsIt uses too many tokens
Question 7hard
What should the case study include? (Select all that apply)
The problem and its cost, and why this scopeResults tying eval scores to a business metric and ROI/paybackHonest limitations and next steps, plus a demo on realistic inputsA line-by-line code walkthrough as the lead
Select all that apply.
Question 8hard
How should the case study be framed?
Technology first, business value as an afterthoughtValue first — lead with the problem and outcome; technology is supporting evidenceOnly the architecture diagramOnly the eval scores
Question 9medium
Why build the riskiest part (e.g. retrieval quality) first?
To make it harderTo learn early whether the slice is viable, before investing in the restBecause it's the most funTo increase cost
Question 10hard
Which are capstone pitfalls? (Select all that apply)
Over-scoping into a broad, half-built platformSkipping evals, guardrails, or deploymentA tech-tour case study that buries the business outcomeDemoing on realistic and hard inputs including an abstention
Select all that apply.
Question 11medium
What is the purpose of the track-to-capstone checklist?
To make the project longerTo verify you actually exercised every part of the course (build, eval, security, deploy, delivery)To choose a modelTo replace the case study
Question 12medium
What portfolio artifact (two words) — not the code — is the real deliverable a client/employer evaluates?
Question 13hard
Complete the capstone's guiding stance: '____ first, technology as proof' (one word).

0 / 13 answered

Capstone: Ship an End-to-End Agent — Practice quiz