Jobhunter is a production MVP for an IT recruitment platform. It is built to show a complete product loop: job discovery, candidate applications, recruiter review, admin operations, release automation, and local production observability.
The project is intentionally portfolio-ready. It does not stop at CRUD screens; it includes role-aware workflows, production guardrails, test coverage, release artifacts, container packages, and an operations stack that can run locally without a public domain.
- Candidates search jobs, compare salary and skills, save opportunities, apply with CVs, manage a CV library, and track resume status.
- Recruiters manage applications inside their company scope, filter pipelines, update statuses, and leave audit notes.
- Administrators manage operational data across users, companies, jobs, skills, roles, and permissions.
- Search-first public job board with city, skill, level, salary, sort, pagination, job detail, top employers, content hub, subscriber flow, and About section.
- Candidate workspace with account-scoped saved jobs, CV URL/upload application, CV library, application history, and status timeline.
- Recruiter workspace with resume pipeline, filters, status updates, audit notes, and company-scoped permissions.
- Admin workspace with operational management screens, pagination/filter patterns, destructive-action confirmation, and clear error states.
- Auth and account flows with login, register, forgot/reset password, HttpOnly cookie auth, RBAC, and email preferences.
- Custom Jobhunter logo mark combining a job briefcase, search lens, and upward career signal.
- Full logo asset for header/footer and compact mark for favicon/auth surfaces.
- Generated demo-company brand images for every seeded employer, including companies that previously had no image.
- Clean red, white, slate, and sector-accent palette designed for an IT recruitment portfolio rather than copied third-party brand assets.
- Visual asset maintenance and screenshot refresh workflow are documented in Visual Assets And Product Screenshots.
- Backend: Spring Boot 4, Java 21, Spring Security, JPA, Flyway, MySQL 8.4, Actuator, Prometheus metrics, Micrometer tracing.
- Frontend: Next.js 16, React 19, TypeScript, TailwindCSS, Jest, Testing Library, Playwright E2E, visual regression.
- Security: production startup guard, unsafe-method client header, in-memory MVP rate limiting, allowlist rich-text sanitizer, upload validation, scoped resume access, and conflict-safe delete behavior.
- Delivery: GitHub Actions CI/CD, GitHub Releases, Docker Hub images, GitHub Container Registry packages, smoke tests, audit checks, and versioned release notes.
- Local production stack: Prometheus, Blackbox Exporter, Alertmanager, local alert webhook, Loki, Promtail, Grafana, OpenTelemetry Collector.
- Staging stack: separate Docker Compose override, ports, network, and volumes for release-candidate checks before local production.
- Data protection: scheduled MySQL backups, manual backup script, restore script, Git-ignored backup folder, and restore rehearsal guidance.
- Observability: backend health/metrics/traces, frontend client error reporting, structured JSON error logs, dashboard provisioning, and local alert log output.
These images are captured from the local production container stack and refreshed through npm run qa:local -- --screenshots.
| Experience | Screenshot |
|---|---|
| Public job board with search, dense job cards, top employers, and content hub |  |
| Job detail with salary emphasis, JD content, company context, and apply panel |  |
| Candidate workspace with saved jobs, CV library, application history, and status timeline |  |
| Recruiter resume pipeline with company-scoped workflow and audit notes |  |
| Admin user management for operational control |  |
| Gemini-powered AI assistant with job-context prompts and live chat response |  |
Jobhunter is suitable for:
- Full-stack portfolio review.
- Internship or job interview demonstration.
- Local product demo without a public domain.
- Foundation for a real recruitment product after external email, object storage, Redis rate limiting, and hosted observability are added.
- In-memory rate limiting is appropriate for a single-node MVP; use Redis before multi-instance deployment.
- Local observability is intentionally self-hosted through Docker Compose; a hosted deployment should use managed log/trace/alert storage.
- Email flows depend on runtime SMTP configuration.
- More destructive admin operations can be expanded into a broader audit ledger later.