Skip to content
All work

Case study

AI Certification Exam Prep Platform

In plain English

Practice for vendor certification exams with realistic question types, timed tests, and a human-reviewed AI question bank.

New certifications drop in as packs without changing the core platform.

Who it is for

  • Certification learners who need original practice aligned to published skills outlines, including interactive item types typical quiz apps cannot render.
  • Content and ops admins who manage packs, generate and review questions, upload assets, and manage users.

Problem

A page of multiple-choice quizzes does not prepare people for real certification exams, and leaked dumps are not a legitimate path.

Quality question banks do not scale by hand. AI can draft items, but they must be scored and reviewed before learners see them.

Solution

We built a web platform where new certifications land as packs of config and assets, not code changes. Learners study with instant feedback or sit timed exams with scaled scores and domain breakdowns. Question types include drag-and-drop, ordering, hotspot, case study, active screen, and simulated labs, not only multiple choice.

An AI tutor is tied to the current question and session. Admins generate drafts by objective, score them for correctness, structure, and clarity, then accept or reject before publish. Subscriptions run through Stripe.

Outcomes

Study, exam, analytics, and admin all run on one Next.js and FastAPI stack, deployed with Docker and AWS ECR.

New certifications and reviewed AI items can be added without changing the core platform.

Next.jsFastAPIPythonSQLiteOpenAIStripeDockerAWS ECRGitHub Actions

How it works

Three layers: learner and admin web app, exam engine, and content plus AI generation.
Three layers: learner and admin web app, exam engine, and content plus AI generation.

Two simple journeys

Learner practice journey and the AI-drafted, human-reviewed content factory.
Learner practice journey and the AI-drafted, human-reviewed content factory.

What we built

Learner product

  • Catalog and enroll

    Browse cert packs, enroll, and resume from a course hub.

  • Study mode

    Instant correct or incorrect, explanation, learn links, and an AI tutor with selectable personas.

  • Timed exam

    Exam chrome with navigator, mark for review, calculator, and chat locked until submit.

  • Nine question types

    Single choice, multi-select, drag-and-drop, ordering, hotspot, case study, active screen, problem-solution, and simulated labs.

  • Results and analytics

    Scaled score on a 100-1000 scale, pass or fail vs pack passing score, domain chart, and weak objectives.

Admin

  • Packs and questions

    Create, edit, and archive catalog entries with quality scores.

  • AI generation

    Draft by objective, type, and difficulty. Accept or reject before items go live.

  • Assets

    Upload media for hotspots and screens; report broken media.

How it works

The Next.js UI is the full product. FastAPI is the exam engine: pack load, domain-weighted selection, grading, scaled scoring, tutor chat, and generation.

New certifications drop in as exam packs (metadata, objectives, seed questions, and optional screen templates) without engine changes. Generation runs in-process with a human accept-or-reject gate.

Integrations

  • OpenAI

    Original question authoring, quality scoring, and a streaming AI tutor.

  • Stripe

    Checkout for paid plans.

  • AWS ECR and GitHub Actions

    Repeatable Dev and QA deploys.

Have a similar challenge?

Start a project