The Ethiopian job market was fragmented across multiple portals with no unified, intelligent search layer. Job seekers faced slow keyword-only searches that failed to surface semantically relevant opportunities. Scrapers were repeatedly blocked by IP bans, and there was no real-time synchronization between the data pipeline and the search engine.
The frontend (HojiiNet) is a Next.js application featuring a hero search experience, job listing page with real-time debounced filtering, company browsing, and user authentication powered by Auth.js. The UI is built on shadcn/ui components with TailwindCSS, supporting both dark and light modes out of the box.
The backend is composed of four decoupled layers: a Tor-proxied scraping layer that harvests jobs from multiple Ethiopian portals; a PostgreSQL persistence layer managed via FastAPI and Alembic migrations; a RabbitMQ message broker that triggers real-time sync events; and a Meilisearch engine configured with hybrid vector search for intelligent semantic relevance ranking.
“We built a system that doesn't just index jobs — it understands them. By combining vector embeddings with an event-driven sync pipeline and a seamless Next.js UI, we gave Ethiopian job seekers the kind of search experience they deserved but never had.”
The result is a production-grade full-stack job platform that runs autonomously — scraping, cleaning, embedding, and indexing thousands of jobs daily with zero manual intervention. The HojiiNet frontend delivers a premium, responsive experience backed by a powerful hybrid search engine, making it the most technically advanced job search platform built for the Ethiopian market.
