Recruitment AI Matcher

Traditional recruitment search is keyword-based: a recruiter types 'React developer 3 years' and gets every CV mentioning those words, ranked by nothing meaningful. Senior recruiters know a great candidate when they read the CV — but they can't read 500 of them per role.

Golden Sea built a platform that approximates that senior recruiter's judgment at scale. Each CV and job description is vectorized using a fine-tuned embedding model and stored in Chroma, a vector database. When a new job is uploaded, the system retrieves the most semantically similar CVs — candidates whose experience and skills genuinely match the role intent, not just the surface keywords.

But semantic search alone isn't enough. Each candidate is then scored across 12 evaluation dimensions: technical skill match, years of relevant experience, industry background, education fit, language requirements, geographic proximity, seniority alignment, leadership signals, domain specialization, project scale, cultural markers in writing style, and upward career trajectory. Each dimension has an individual score; the aggregate produces a match percentage.

Candidates are auto-sorted into three tiers: Top Match (recommended for interview), Average (second look), and Poor (no fit). Recruiters see a visual breakdown per candidate explaining why they ranked where they did — not a black-box score, but a transparent reasoning trail. The platform integrates directly with the agency's existing CV database and LinkedIn sourcing workflows.