01 / Hackathon project
Eban unified hiring-analysis database
Open the live siteThe idea
(01)Job ads say a lot about a company: what it builds, which specialists it needs and where its plants are. Eban collects vacancies from 16 Russian job sites into one base, ties every vacancy to its employer by tax number, enriches each company with its state contracts, procurement records and contacts, links companies into holdings in one graph, and checks each of them for links to the defense industry.
The result is a searchable map of the industry. Out of 258k employers, 4,642 show defense signals, and 2,595 of them are confirmed: 2,469 are in the official defense-industry register, 72 are under defense sanctions and 54 have three or more independent signals.
How it works
(02)- 01
Collect
Scrapers pull vacancies from 16 job sites, with salary, location, skills and employer details.
- 02
Identify
Every vacancy is tied to its employer by tax number (INN), so one company gets one profile across all sites.
- 03
Enrich
For each company the pipeline pulls its state contracts and public procurement records, and parses contacts, including the names and surnames of employees and managers.
- 04
Build the graph
Companies are grouped into holdings by shared owners, phones and emails, and merged into one graph of connected companies.
- 05
Classify
Registers and sanctions lists give hard evidence. A language model reads company data and assigns a defense direction with a quoted proof and a confidence level.
- 06
Explore
A region map, a filterable company list, a rating, the company graph and a card for every company with its vacancies, contracts, contacts and evidence.
Overview
(03)Secrethon 2026
(04)Eban was built in a team at Secrethon 2026, a hackathon by AI Meetups with monobank and the Ministry of Defence of Ukraine as partners. The team reached the final of the Ministry of Defence track.