Energy Recruitment Hub

Project overview

Your role

• Full stack development across Strapi 5 backend and Next.js 16 frontend
• Backend: custom auth, two layer RBAC, Stripe billing with webhooks, OpenAI integrations, S3 CV uploads, PostgreSQL
• Frontend: App Router architecture, Redux state, Tailwind v4 and shadcn/ui, SEO and Core Web Vitals optimization
• Custom Strapi content types, controllers, services, and permission bootstrap
• Recruiter subscription billing, job lifecycle approval workflow, and applicant unlock flows
• AI CV checker, career intelligence, interview prep, and AI assisted job drafts
• BFF auth routes with HttpOnly JWT cookies and three layer route gating
• Sentry observability and documentation driven feature workflow

Tech stack used

• Frontend: Next.js 16, React 19, Tailwind CSS 4, shadcn/ui, Redux Toolkit, TipTap 3, Framer Motion
• Forms & HTTP: React Hook Form, Axios (JWT cookie + localStorage fallback), lucide-react
• Backend: Strapi 5 (TypeScript), @strapi/plugin-users-permissions, PostgreSQL
• Payments & AI: Stripe 17 (Checkout, Portal, webhooks), OpenAI 4 (CV scoring, job drafts, career intelligence)
• Storage & Parsing: AWS S3 (CV uploads), pdf-parse, mammoth (DOCX), xlsx (bulk import), fuse.js
• Observability: Sentry (@sentry/node, @strapi/plugin-sentry), nodemailer
• SEO: next-sitemap, JSON-LD, canonical domains, PWA manifest

Key features

• Public job board with search, filters, split pane layout, and direct to recruiter applications
• AI CV checker with deterministic rule based matching and LLM relevance gate
• AI interview prep with per job cached questions and career context prep
• Career Intelligence: role matching, skills gap analysis, salary insights, career journey dashboard
• Recruiter Stripe subscriptions with 4 plans, usage limits, publish gates, and Customer Portal
• Job lifecycle workflow: draft, pending, active with admin approve/reject and billing gates
• AI assisted job drafts and Excel bulk job import
• Multi role dashboards: Admin, Recruiter, VA, and Job seeker with scoped pages
• Auth: email/password, Google OAuth, OTP, email verification, persona switching
• Companies, Projects, and Training modules with ownership rules
• SEO and PWA: structured data, sitemap, robots, web manifest, llms.txt

Platform modules

User roles

Target audience

• Energy sector job seekers seeking roles, CV feedback, and career guidance
• Recruiters and employers posting jobs and unlocking applicants via paid plans
• Virtual assistants moderating jobs, companies, and training content
• Platform admins managing users, approvals, and system settings

Code structure & architecture

job_portal_cms/  (monorepo)
├── backend/                    # Strapi 5 (TypeScript)
│   ├── src/api/                # 23 API route groups
│   ├── src/services/           # recruiter-billing, ai-cv-matcher, stripe/*
│   ├── src/middlewares/
│   └── src/index.ts            # Bootstrap: roles, permissions, patches
└── frontend/                   # Next.js 16 App Router
    ├── app/                    # 99 pages, role prefixed segments
    ├── app/api/auth/           # BFF layer (HttpOnly strapi_jwt cookies)
    └── store/                  # Redux Toolkit slices

Request flow:
Browser → Next.js → BFF /api/auth → Strapi REST /api → PostgreSQL
                              ├→ AWS S3 (CVs)
                              ├→ Stripe (billing)
                              └→ OpenAI (AI features)

Performance optimizations

• LCP fixes: optimized next/image with responsive sizes on hero and banner assets
• Home page below the fold sections moved to next/dynamic lazy loading
• Google Analytics deferred via next/script strategy afterInteractive
• System font stack (font-sans), no render blocking next/font
• Core Web Vitals targets: Performance > 80, LCP < 2.5s, CLS < 0.1, INP < 200ms
• AI hot path uses deterministic CV scoring; OpenAI gated behind relevance check
• Interview prep questions cached per job and reused for later applicants
• NODE_OPTIONS max old space size cap in production start
• Usage counters synced from DB after publish to avoid drift
• Mobile responsiveness audit: tablet nav, body scroll lock, touch targets

• Thin controllers, fat services: domain logic in backend/src/services/
• One way dependency chains: resume-detection → cv-relevance-gate → ai-cv-matcher
• Two layer RBAC: canPerform(securityRole, persona, action) requires both layers
• Ownership stamping: created_by_user_id immutable from request body on update
• Strapi 5 draft/publish dedupe helpers (dedupeJobsByDocumentId) on list endpoints
• Redux Toolkit slices: auth, jobs, companies, recruiter, jobSeeker, admin, projects
• Three layer auth: Next.js middleware → Strapi permissions → BFF cookie handling
• Documentation driven development with feature docs in docs/features/*

Challenges & solutions

• Strapi 5 documentId duplication from draft/publish creating multiple rows.

  Shared dedupe helpers (dedupeJobsByDocumentId, dedupeRecruiterJobsByIdentity) on all list endpoints.

• Two authorization needs: hard role ceiling vs UX persona.

  Two layer RBAC with role.type for security and user_type persona; canPerform() requires both.

• Recruiter and job seeker switching without re registration.

  Persona switch updates user_profiles.user_type only, keeping Strapi security role intact.

• Stripe billing complexity with limits, usage, and publish gates.

  Strapi owns limits and gates; Stripe owns subscriptions; webhook sync with idempotency table.

• Irrelevant or non resume uploads skewing AI scores.

  detectResumeLikeDocument and evaluateCvRelevanceGate run before deterministic scoring.

• Circular service imports across CV and Stripe modules.

  Refactored into one way dependency chains without behavior change.

• Core Web Vitals on image heavy marketing pages.

  Image optimization, dynamic imports, and deferred third party scripts.

• SQLite to PostgreSQL migration.

  Postgres only config, Docker compose, import script, and removal of SQLite artifacts.

Results

• Production grade platform with 99 frontend pages and 23 backend API modules at energyrecruitmenthub.com.
• Monetization ready with 4 Stripe subscription tiers, usage limits, publish gates, and Customer Portal.
• AI differentiation: CV scoring, career intelligence, job drafts, and cached interview prep with cost controlled LLM usage.
• Secure role aware system with three layer auth, frozen permission matrix, and ownership enforcement.
• SEO and performance optimized: Core Web Vitals targets met, structured data, canonical domains, PWA manifest.
• Maintainable and observable with Sentry, documentation sync workflow, and audit reports.