birjob.com

Azerbaijan's job market is unusually fragmented for its size. A candidate looking for a software role in Baku has to check half a dozen general boards, the careers pages of every bank and telecom, a rotating cast of recruiter Telegram channels, and a handful of category-specific platforms. Each one has its own UX, its own filtering quirks, and its own definition of what counts as an active vacancy. Listings expire silently. Identical roles get cross-posted to three different boards under three different titles. Salary information is missing from most of them.

The cost of that fragmentation falls on candidates, who burn hours every week re-checking the same five tabs, and on employers, whose roles are invisible to anyone who didn't happen to land on their specific board that day. birjob.com exists to collapse that surface area into a single, free, no-account feed of every active vacancy in the country, refreshed daily, normalised into a consistent shape, and searchable in one place.

The core surface is the listings feed: more than 4,000 vacancies on a typical day, drawn from over 50 sources and refreshed every 24 hours. Visitors browse without an account, filter by category, drill into a company's page to see every role that company is currently hiring for, or jump straight to the original posting on the source site.

The discovery layer is built around three primitives. Categories normalise listings across sources — a backend role posted as 'Software Engineer' on one board and 'Backend Developer' on another resolves into the same IT/Software bucket. Company pages aggregate every open role per employer, surfacing scale signals like 'Andersen — 128 active vacancies' or 'ABB — 104 active vacancies' that tell candidates which firms are actually hiring at volume right now. Pagination keeps page weight bounded even when the index swells past 8,000 rotating entries.

Around the listings feed, two content surfaces compound traffic over time: a 2026-tuned salary calculator covering net ↔ gross ↔ supergross conversions in line with the latest Azerbaijani income-tax brackets, and a career-resources blog covering CV writing, interview preparation, motivation letters, and a library of frequently-asked interview questions. Both are free, both are localised, and both serve to keep candidates returning between active job searches.

Aggregating fifty sources into one trustworthy index is not a single problem — it is six problems stacked on top of each other, and each one has to be solved well enough that none of the others becomes the bottleneck.

The first is scraping cadence. Hit every source every five minutes and you get fresh data plus an angry hosting bill plus an eventual block. Hit every source once a week and you become irrelevant. A daily cadence per source is the equilibrium most aggregators settle on: fresh enough that no listing is more than 24 hours stale, polite enough that source operators tolerate the traffic, cheap enough that the unit economics work on advertising revenue alone.

The second is per-source resilience. Any one of fifty sources can change its HTML overnight, return a captcha wall, or simply go down. The aggregator has to absorb that without losing the other forty-nine. The shape that works is one isolated scraper per source, each one independently retried and independently logged, with the index treated as eventually-consistent rather than transactional.

The third is deduplication. The same role gets cross-posted to three boards. A naive index treats those as three vacancies; a useful index treats them as one. Some combination of fuzzy title matching, employer normalisation, and posting-window proximity collapses the duplicates without merging genuinely distinct roles that happen to share a title.

The fourth is normalisation: titles, locations, employment types, and categories vary wildly across sources, and the canonical taxonomy on the aggregator side has to be stable enough that filters keep working as new sources are added. The fifth is expiration: most sources don't reliably mark listings closed, so the aggregator has to infer it from absence on the next crawl. The sixth is the front-end discovery layer that turns the index into something a candidate actually wants to use — fast pagination, useful categories, and the company-directory cut.

Aggregation alone is a commodity. Every job board wants to be the one place candidates go, and a candidate has no particular loyalty to a feed that isn't differentiably better than the next one. The differentiator on birjob is the surrounding content: the salary calculator and the career-resources blog.

The salary calculator matters because Azerbaijan's net / gross / supergross conventions are confusing even to people who deal with them daily. Recruiters quote net, employers post supergross, payroll calculates gross, and the brackets shift year to year. A localised, current calculator that handles every direction of conversion in real time is genuinely useful, and it brings candidates back to the site between active job searches — exactly the engagement pattern that drives long-tail SEO.

The career-resources blog plays the same role on the discovery side. CV writing, interview preparation, motivation-letter templates, common interview questions — these are the queries candidates run on Google long before they're ready to apply. Ranking for them feeds the listings funnel two and three months upstream of an active search.

The "Elan Yayımla" path lets companies publish directly into the feed without going through a third-party board. For employers, that's a free distribution channel into a candidate base that already arrives looking specifically for jobs — higher intent than any social or display channel can offer. For the platform, direct publishing reduces dependency on upstream sources: every directly-published role is one fewer scrape, and one more piece of content the platform owns end-to-end.

The natural next layers are alerting and matching. Saved searches with email or push notifications turn the aggregator from a thing candidates check into a thing that surfaces relevant roles automatically. Salary-aware filtering — once enough listings carry parsed salary data — makes the index meaningfully more useful than the source boards it pulls from. AI-powered match scoring against an uploaded CV is the obvious endpoint, and the kind of feature that converts a free aggregator into something candidates pay for, employers sponsor, or both.