A 102-question diagnostic, a mock interview coach, and a CV that fixes itself. Here's what I learned building AI Career during my studies at SITE, an AI-native university in Geneva.
I still remember the exact moment the idea clicked. A classmate showed me her CV, three different versions, none of them tailored to anything in particular, and asked me which one "sounded more employable." Neither of us knew. We had grades. We had a few projects. What we didn't have was any real signal connecting what we were learning to what employers actually wanted from us.
That gap — between academic performance and career readiness — became the starting point for AI Career, a project I built as part of my studies and collaboration with SITE, an AI-native online university based in Geneva, Switzerland. It started as a coursework idea. It became something I now think every university should have.
The Problem: Career Services Weren't Built for How Students Actually Search
Traditional career centers work on a simple model: post the vacancy, let students browse, hope for a match. It's a static, manual process, and it puts the entire burden of self-assessment on the student. You're expected to somehow know your own strengths well enough to filter through hundreds of listings and figure out which ones you're actually qualified for.
Most students can't do that. Not because they lack ability, but because nobody ever gave them a structured way to translate coursework and personality into career direction. So they either apply to everything, or they apply to nothing.
I wanted to build the layer that was missing: a system that understands the student first, and only then shows them the market.
Starting With the Student, Not the Job Board
The core of AI Career is a career preferences diagnostic — 102 questions, about 17 minutes, built around eight disposition scales: things like conscientiousness, adaptability, collaboration, and grit. Instead of asking students to self-report ("Are you a leader?"), it uses paired comparisons — for each block of statements, you pick what fits you most and what fits you least. That forced-choice format produces a much more honest signal than a standard survey, because you can't just rate yourself highly on everything.
At the end, the system maps you to an archetype — mine came back "Organizer," with "Catalyst" as a runner-up pattern — and shows your full profile as a shape across all eight scales, so you can see at a glance where you actually stand out versus where you're simply average.
That profile becomes the input for everything downstream: which vacancies get surfaced, which skills get flagged as gaps, and what the AI recommends you work on next.
AI Career Matching, Not Just Job Listings
Once a student has a profile — diagnostic results, academic record, stated interests — the platform matches them against live vacancies and shows a percentage match for each one, along with salary range, work format, and role tags like entry-level or fast-paced. It's a small design choice, but it changes the psychology of job searching completely: instead of scrolling a wall of listings wondering if you're even qualified, you see your fit score before you decide whether to invest time in an application.
This is the same idea behind SITE's broader AI career matching engine, which continuously aligns student learning with live employer requirements — AI Career was, in effect, the practical, hands-on version of that concept, built by a student, for students.
Where the AI Actually Does Work: CV, Cover Letters, and Interviews
The diagnostic is the foundation, but the part students seem to like most is the AI assistant layered on top of the profile. From "My Profile," you can ask it to optimize your CV against actual feedback, generate a cover letter tailored to a specific vacancy pulled directly from the platform, or plan your portfolio — mapping which projects and certificates would actually move the needle for your target role.
There's also an AI mock interview module, so instead of walking into a real interview cold, students can rehearse against a simulator that responds to their answers, and a technical case simulator for roles that expect you to demonstrate applied skills rather than just describe them on paper.
None of these are generic AI wrappers bolted onto a job board. Each one reads from the same underlying student profile, so the CV advice, the interview prep, and the vacancy matches are all pulling from the same data instead of treating career prep as five disconnected tools.
Why I Built This at a University, Not a Startup
I built AI Career as a student project during my studies at SITE, an AI-native online university registered in Geneva, Switzerland, where the curriculum is built around applying AI to real problems rather than just studying it in theory. That context mattered more than I expected. Being inside a university, with direct access to actual student profiles, actual course data, and actual questions students were asking their advisors, meant the product could be shaped by real friction instead of assumptions about what students "probably" need.
It also meant the project didn't stay a private experiment for long. It was originally built for SITE's own students, but the architecture — diagnostic engine, matching layer, AI career assistant — isn't specific to one institution. It's now open for collaboration with other universities and companies that want to give their own students or candidates the same kind of structured, data-driven career support, instead of a static list of job postings and a PDF template for CVs.
What I'd Tell Any Student Thinking About Building Something Similar
The instinct when you set out to build a "career platform" is to start with the job listings — scrape vacancies, build filters, ship it. I'd argue that's backwards. The listings are the easy part. The hard part, and the part that actually creates value, is building an honest picture of the person first. Once you have that, matching, CV help, and interview prep all become downstream features instead of separate products you have to sell independently.
I also learned that forced-choice diagnostics (pick the one that fits most, pick the one that fits least) produce much better signal than open self-rating scales. Students are bad at rating themselves in isolation. They're much better at comparing.
Where AI Career Goes From Here
Right now, AI Career lives as a working prototype built and tested within SITE. The next steps are the obvious ones: more employer partnerships feeding live vacancies into the matching engine, deeper integration with academic records so the diagnostic updates as students actually grow, and validating the approach with more universities outside SITE's own student base.
If you're a student building something in this space, or a university or company curious about integrating a diagnostic-first career matching engine into your own programs, I'd genuinely like to hear from you.
About SITE
SITE is an AI-native online university registered in Geneva, Switzerland. We offer accredited BBA and MBA programs powered by a proprietary AI career matching engine, gamified learning platform, AI-driven content personalization, and access to an international exchange network across 20+ partner institutions. Accredited by QAHE and EQAC, ISO 21001:2018 certified, CEEMAN and ECBE member. 1,200+ students enrolled from 20+ countries since September 2024.
Learn more at sitegeneva.com · info@sitegeneva.com · Chemin Louis-Hubert 2, 1213 Petit-Lancy, Geneva, Switzerland
Top comments (0)