Recruitment / AI

AI Recruiting Platform

Varrlyn is a boutique Amsterdam consultancy that places specialists at financial institutions. Hidden Harvest built them an end-to-end recruiting platform: it scrapes and standardises vacancies, turns every CV into a searchable vector, and surfaces the strongest candidate matches per role — so the team spends its time on people, not portals.

Varrlyn

Five profile clusters

Synthetic demo data, no real CVs.

2,668

CVs vectorised & searchable

217

Live vacancies across 3 portals

Sourcing portals automated

Figures as of July 2026.

01 · The Challenge

Matching ran on manual labour

Varrlyn sources contract specialists from multiple vacancy portals, then matches them against a large pool of CVs. Done by hand, that meant hours of portal-hopping, inconsistent vacancy data, and a real risk that the best-fit candidate never surfaced — because no recruiter can hold thousands of CVs in their head.

02 · The Vision

One pipeline, from portal to shortlist

The goal was a single platform that gathers and standardises vacancies automatically, structures and understands every incoming CV, and computes candidate matches on demand — turning a manual sourcing grind into a ranked shortlist the recruiter can start from.

Around the table with the Varrlyn team, reviewing the work on the meeting-room screen

Walking the Varrlyn team through the platform, a vacancy up on screen

03 · The Solution

Scrape, standardise, vectorise, match

Hidden Harvest built the full path end to end. An async scraper pulls vacancies from the sourcing portals, Claude standardises and filters them into a clean model, and each is stored in Postgres. In parallel, CVs are structured, embedded and projected into a pgvector search space — visualised as a live 3D "galaxy" of candidates. For any vacancy, the platform computes and ranks the closest matches on demand, right in the dashboard.

04 · The Result

Thousands of CVs, one query away

The platform runs live today: vacancies flow in and are standardised automatically across three portals, and thousands of CVs are structured and searchable by meaning rather than keywords. Instead of hunting through portals and folders, the team starts from a ranked shortlist.