Skip to content
Part of ELO — Embedded Logic Operations · AI solution demo lead & engagement sponsor ↗

Live since July 2025

Vimalgo

AI-driven headhunting platform for a Swiss recruitment firm — CV parsing, candidate profiling, and job-candidate matching with end-to-end application tracking. Built from August 2024 and live since July 2025.

Role
AI solution demo lead & engagement sponsor (via ELO)
Period
Aug 2024 — present · live since July 2025
Stack
React · Node.js · TypeScript · AI CV parsing · and 4 more
Place
Cannella & Partner Solutions AG · Möhlin, Switzerland (Remote)
  • Cannella & Partner Solutions

    The AI demo that got them to bet early on AI. Their headhunting platform has been live since July 2025.

Track Application — applicant information, staged pipeline (started → interviews → trial → result), consultant feedback notes and document upload.

Track Application — applicant information, staged pipeline (started → interviews → trial → result), consultant feedback notes and document upload.

July 2025Live · in production

A three-person recruitment firm competes on how quickly it can put the right person in front of a client. Every hour a consultant spends reading CVs, retyping details into a profile, and re-scanning job portals is an hour not spent on the placement itself — and in 2024 the credible version of "AI will fix this" was still mostly a promise.

01 / 03

The bet

Cannella first planned a traditional CV-sorting system.

Fauzul took the risk of building an AI demo for their pitch instead, and led the demo that made the promise concrete: an AI-driven matching flow the client could try against their own material rather than a slide deck about what it would eventually do. That demo is why the client committed early instead of waiting for the category to mature, and the platform went live in July 2025.

02 / 03

What the platform does

  • AI-powered CV parsing. Résumés become structured candidate profiles — personal summary, skills, experience, certificates, degrees, fields of study — without retyping.
  • Job–candidate matching. Automated matching surfaces high-potential pairs between the candidate pool and the live job database with minimal manual work.
  • Self-updating job database. Job records auto-update with new posts pulled from different job portals, so the pool being matched against stays current.
  • Advanced search and filtering. Narrow a candidate pool down to a shortlist for any specific job across skills, study fields, degrees, certifications, and tags.
  • End-to-end application tracking. Every application moves through an explicit pipeline — started, interview rounds, trial, result — with per-stage feedback notes and document uploads attached to the record.
03 / 03

Why it holds up

The AI is the wedge, not the whole product.

Parsing and matching only pay off if the surrounding system is honest about state — who was interviewed, by whom, what they said, what happens next — so the platform was built around an explicit application pipeline with feedback and documents attached at each stage, rather than an inbox and a spreadsheet with a model bolted on.

That is also what made the early bet safe: the recruitment workflow keeps working on its own terms, and every improvement to the matching quality lands on top of it.

Decisions and why

  1. Demo the AI flow before scoping the platform

    WhyThe client's decision was whether to trust AI matching at all. A working demo against their own material answered that question in a way a specification could not, and set the scope of everything built afterwards.

  2. Structured profiles as the matching substrate

    WhyMatching quality is bounded by what the system knows. Parsing CVs into explicit skills, experience, certificates, and study fields gives the matcher — and the consultant reviewing it — something inspectable to reason over.

  3. Application pipeline as first-class state

    WhyRecruitment fails on lost context, not on shortlisting. Modelling stages, feedback, and documents on the application record keeps the human process auditable while the AI handles the volume work.

Technical highlights

  • 01

    AI CV parsing into structured candidate profiles (skills, experience, certificates, degrees, fields of study).

  • 02

    Automated job ↔ candidate matching over a continuously updated job database.

  • 03

    Job-portal ingestion keeping the job database current without manual entry.

  • 04

    Explicit application pipeline — started → interview rounds → trial → result — with per-stage feedback and file attachments.

  • 05

    Multi-language UI for a Swiss client operating across languages.

Impact

  • Live since July 2025 and running as the client’s recruitment system — the early AI bet is delivering.
  • Removed manual CV data entry from the consultant workflow via parsing into structured profiles.
  • Turned a three-person firm’s candidate pool into a searchable, matchable asset rather than a folder of documents.
Outcomes

By the numbers.

AI parsing
CV → structured profile
Matching
Job ↔ candidate, ranked
Swiss client
Cannella & Partner Solutions AG
Visit project

See it in the wild.

This case study covers how Vimalgo was built — these links open what ships today: the live product, repo, demos, and supporting material.

↗ Next case study

TheBestBroker

TheBestBroker helps investors pick a broker instead of guessing at one. I led the project through 2023 and 2024: broker listings and comparisons with an affiliate system attached, and on top of it an AI chat advisor that works off a dynamic form — the user describes what kind of investor they are and what they want to trade, and the recommendation follows from that rather than from a generic ranking. The interesting constraint was the platform: the client required the site to stay in Webflow so their own team could keep editing it, so everything intelligent had to be built on top of Webflow rather than instead of it.