RecruitSecure AI — semantic candidate search

What it is

RecruitSecure AI (recruitsecureai.com) is a SaaS product for semantic candidate search whose AI runs 100% locally, inside the browser: zero external APIs, zero data sent anywhere — privacy and GDPR compliance by design. It started as the CV Finder I built at UB Manufacturing, and I took it from prototype to a shipped product of my own.

The problem it solves

Keyword screening misses synonyms and context: “React” is not the same as “frontend with IoT experience”. RecruitSecure understands intent and meaning — you search in natural language and it ranks candidates by real semantic relevance, not exact term matches.

How it’s built

• AI in the client: vector embeddings with Transformers.js (Hugging Face) running in the browser. CVs never leave the recruiter’s device → total privacy and no per-API cost.
• Never blocks the UI: embedding computation runs in dedicated Web Workers; the interface stays fluid while it processes.
• Memory under control: LRU embedding cache + IndexedDB persistence to handle large CV volumes without overwhelming the browser.
• Two-phase search: cheap boolean filters first (location, years), then embeddings only over the resulting subset → fast answers on complex queries.
• Robust extraction: text pipeline for PDF/DOCX/TXT with technical-term detection to sharpen the matching.

A product, not a prototype

It is live as a SaaS, with Pricing, Security and API pages. It shows the full journey: spotting a real problem at UB Manufacturing, building the solution, and turning it into a published product of my own.

Stack: Next.js · React 19 · TypeScript · Transformers.js · Tailwind CSS.