Case study · 2026

Sky Resume — AI Resume Builder SaaS

Resume and cover-letter SaaS with AI writing, 10 ATS-friendly templates, PDF export and Stripe subscriptions — MVP in about two months.

  • Next.js
  • Django
  • Django REST Framework
  • PostgreSQL
  • OpenAI API
  • Stripe
  • JWT
  • Docker
  • next-intl
  • Zustand
Sky Resume system design overview

Context

A startup building an AI-powered resume platform, with a roughly two-month window from concept to production.

Constraints

  • AI features must be useful and cost-controlled per user
  • The PDF must look exactly like the live preview
  • Editing must never lose work
  • Free and paid plans from day one

System design

            User
             │
   ┌─────────▼──────────┐
   │ Next.js client     │  editor, drag-and-drop sections,
   │ (Zustand, RHF+Zod) │  live preview, autosave, i18n
   └─────────┬──────────┘
             │ REST + JWT (Google / Facebook / email login)
   ┌─────────▼──────────────────────────────────────────┐
   │ Django REST API                                    │
   │  accounts · resumes · cover_letters · ai_core ·    │
   │  billing (plans, entitlements, quotas)             │
   └──┬──────────┬───────────┬────────────┬─────────────┘
      ▼          ▼           ▼            ▼            ▼
  PostgreSQL  OpenAI API  Stripe API  PDF conversion  SMTP
                          (+webhooks)  service        (verification)

AI request flow

Frontend ──POST /api/ai/rewrite (text, tone)──► AIViewSet
AIViewSet: check plan quota ──► ai_service.rewrite_section()
ai_service ──prompt──► OpenAI ──rewritten text──► ai_service
ai_service ──log usage──► AIUsageLog (DB)
AIViewSet ──200 { rewritten_text }──► Frontend

Key decisions

  • Decoupled frontend and API so web, admin and future mobile clients share one backend.
  • Entitlements as data. Plans map to feature limits (e.g. free = 2 previews/month, pro = unlimited) in the database — pricing changes need no deploy.
  • Quota check before every AI call, usage logged after — cost is bounded and auditable per user.
  • Stripe webhooks as the source of truth for subscription state, not client redirects.
  • Autosave so edits are persisted continuously and never lost.

Delivered

  • Real-time editor with drag-and-drop section ordering
  • AI summaries, bullet points, keyword optimisation and cover letters
  • 10 ATS-friendly templates; PDF export consistent with the preview
  • Email verification, social login, multi-resume dashboard, admin analytics
  • Dockerised dev and production environments

My role

Full-stack developer (architecture, frontend, backend, AI integration, deployment), working with a project manager — delivered through KohzadTech.

More projects

  • Dubai Haus website homepage
    2025–26Case study

    Dubai Haus — Real-Estate Platform, SEO & Growth

    Bilingual property platform built from scratch and grown through SEO — 2,200+ synced projects and 6.8× more organic clicks in 12 months.

    • Next.js
    • PostgreSQL
    • Prisma
    • Redis
    • +8
    Read case study
  • Saadiyat Marina website homepage
    2026Case study

    Saadiyat Marina — Bilingual Property Publication

    German-first Next.js publication on Abu Dhabi's Saadiyat Island — 54 URLs with a complete hreflang cluster, built in 7 working days.

    • Next.js 15
    • TypeScript
    • JSON-LD
    • hreflang
    • +3
    Read case study
Rehan Khan face Picture

Let's work
together

Version
2026 © Edition
Local time
Get in touch