Talent1

The problem

Keyword search over a candidate database misses strong matches whose résumés use different vocabulary than the job description. Compounding that, API latency made every recruiter interaction feel sluggish.

What I built

Refactored the platform across Next.js and Express, then introduced vector-embedding semantic search so candidates surface by meaning rather than exact keyword overlap.

Impact

API response times dropped 45%, and recruiters started surfacing candidates that keyword search had been silently missing.

Architecture

Next.js frontend against an Express API. Candidate and role documents are embedded and compared in vector space, replacing the previous keyword-only matching path.

Key features

• Vector-embedding semantic search across the candidate database
• Refactored Next.js and Express architecture
• Latency-focused API optimization