Job search

A search that starts from your profile, not a keyword

Instead of typing keywords, you pick the resume version (base or adapted) you want to search with. It's that resume's actual content, summary, experience, skills, that drives the search.

A first AI step reads that resume and extracts a multilingual search profile: equivalent job titles in French, English, and depending on your region, Portuguese or Arabic, a list of required and optional skills, keywords to exclude, and an estimated experience level (junior, mid, senior, lead). That's what lets a resume written in French surface offers published only in English.

Where the offers come from

CvCy Talent queries several job sources in parallel (general job boards, remote-work boards, humanitarian and NGO listings, and a pool of selected companies), with a caching system to avoid hitting those sources on every single search. Results are then deduplicated: two identical postings published on two different sources are merged into one, keeping the more complete description and the most recent date.

How the relevance score is calculated

Every offer gets a score from 0 to 100, computed through explicit rules, not a black box: job title match, skills shared with your profile, location, language and experience-level match, posting freshness. An offer whose title clearly has nothing to do with your profile, or one posted more than two months ago, gets its score heavily penalized rather than being silently dropped without explanation.

Once that ranking is established by the rules, the top offers are sent to the AI for a fine adjustment: it can nudge the score up or down by a few points and explain why, but it doesn't reinvent the ranking from scratch. The AI refines an already-computed score, it doesn't replace the matching logic.

When criteria are widened, you're told

If a strict search doesn't return enough results, CvCy Talent progressively widens the criteria, first to the region, then to the whole continent, then to remote offers accessible from Africa. Every widening step is flagged, so you understand why some results are less targeted instead of discovering irrelevant offers with no explanation.

Tracking every application in a pipeline

Once you've spotted an offer, you can add it to a visual kanban-style pipeline organized by stage: to apply, sent, follow-up, interview or test, offer received, closed. Every card keeps your notes, the job description, and the salary, so you never lose track of an application among dozens of others.