Helping Stelloa bring AI into the job search

The brief

Stelloa matches developers to jobs using an AI ranking algorithm. Developers get a curated feed. Companies get pre-qualified candidates without sourcing them. When the founders approached me, they had a working algorithm and a trial version with no design involvement. They needed a real product ready for Tech BBQ, where it would be offered to conference attendees for free. I had ten days to deliver so they could implement the designs in time for the conference.

I had to design both sides of the marketplace simultaneously. A mobile experience fast enough to browse between conference talks, and a desktop tool thorough enough to replace a recruiter's spreadsheet.

Information Architecture

Mobile App

A job listing typically runs 500–800 words. I had to identify and fit the key parts into a single swipe card, which the founders referred to as "Tinder, but for jobs". Candidates could sign up through LinkedIn, pulling their profile data to skip manual onboarding. From there, they'd swipe through job cards, each one compressed to the few details that actually drive a decision.

Their early trial version showed company name, role title, a short description, and location. I cleaned up the layout, but the real question was information hierarchy. With desk research & conversations with my students at Redi School to draw on, I made calls based on a simple question:

What signals do people look for when deciding if a job application is worth their time?

I prioritized position, salary range, remote policy, company name & short summary on the card surface. The full description moved to the detail view, where it helps candidates who are already interested learn more. It's useful once someone is already interested, but it's not what creates the interest.

The company side

The desktop dashboard gave hiring managers a pipeline view of algorithm-matched candidates, with enough context to evaluate a match without opening a separate tab.

The desktop application presented different opportunities. Hiring managers receiving algorithm-matched candidates needed to evaluate them efficiently without the overhead of a full ATS. Because candidates were pre-filtered by the AI algorithm, the dashboard could be focused on reviewing quality matches rather than managing volume. So the UI needed to be transparent about the algorithm's decision making and allow for human override.

1/6 Each role gets a pipeline view. New, 1st Interview, 2nd Interview — so hiring managers can track progress without a spreadsheet.

Systems Design & Design System

Stelloa had no brand identity at the time of the project. They knew they'd develop one eventually, but they needed to ship first. I built a design system with token-based styling, which would make integrating a new visual identity later quite simple by enabling them to quickly apply their new brand colors and typography on top of it.

Since then, they've developed their brand and reskinned the app without touching the system design or the information architecture.

1/3 Button components. Variants, states, and sizes defined as reusable tokens — ready to reskin when the brand landed.

Launch

The app launched at Tech BBQ Copenhagen, where it was offered free to attendees and companies. Developers used it between talks to browse real listings. The venue was full of developers actively thinking about their next move, surrounded by companies in need of talent. Ideal conditions for a job-matching app. The feedback was consistent: people found it intuitive and fast. The card hierarchy worked the way I'd intended it to.

The launch and feedback confirmed the hypothesis that developers would barely glance at the description. They were making decisions in seconds based on salary, tech stack, remote policy, and company credibility. Even though it wasn't the focus, the UI got really good feedback from the users, who highlighted how intuitive and fast they found the app. Areas of improvement included the certain bugs and the need for more job listings.

"Love the app and it's aesthetic."

— App Store review, initial release (before the brand update)