Vacancy Index
Allowed sources enter ingestion, normalization, taxonomy, deduplication and freshness gates before a vacancy reaches the catalog.

A specialized HRTech platform that joins a current Vacancy Index, structured Talent Card and Talent Spec, and an explainable matching architecture across five deep-tech verticals.
Vacancies repeat across sources, close without notice and use incompatible job titles. A static PDF does not express depth of skill, production responsibility or evidence of experience.
Keyword search cannot reliably distinguish a passing mention from deep production experience. A single opaque AI score adds false certainty instead of explaining a decision.
Allowed sources enter ingestion, normalization, taxonomy, deduplication and freshness gates before a vacancy reaches the catalog.
Professionals and employers receive mirrored structured models instead of two incompatible free-text documents.
Must-have coverage, strengths, gaps, quality of data and reason codes are shown separately from the final human decision.
Allowed source
Raw snapshot
Normalize + taxonomy
Deduplicate + freshness
Vacancy Index
A verified representative can claim a prepared company asset built from permitted public vacancies, receive a workspace and take ownership of direct vacancy workflow. Candidate data remains private until explicit consent.
Talent Card and Talent Spec are designed as versioned JSON that their owners can export into compatible HR systems. Monetization belongs in workflow and results, not in blocking access to a person’s own data.
I formed the 4ITX product model, built the Russian Vacancy Index and designed the path from disconnected prototypes to a two-sided production platform: ingestion, taxonomy, Talent Card, Talent Spec, explainable matching, employer commerce, privacy, state machines and a vertical-slice delivery roadmap.
Show the task. I will outline the architecture and the boundaries of a useful pilot.
Send a task