quarksUpRolling out

Accelerating recruitment with AI

Role

Product Designer

Timeline

Ongoing

Team

1 Product Designer2 Developers1 Software Engineer1 Product Manager

Skills

AI UXUser ResearchDesign System

Overview

Introducing AI where time is lost the most

quarksUp is an HRIS platform used by more than 500,000 users. The team wanted to introduce AI into the recruitment module to reduce the burden on recruiters. Pre-screening applications, done entirely by hand, took up a significant share of their time. The goal: automate CV review so recruiters could focus on the profiles that matter.

My role

I owned the end-to-end design of this feature: research, problem framing, interface design, and post-launch iteration. Embedded in the product squad, I led the contextual interview sessions, ran design sprints with the team, and worked directly with engineering to define a consistent and actionable scoring logic. The feature is currently in beta deployment with several enterprise clients.

Discover

Understanding how recruiters screen CVs

Define

What the testing showed

Design

Designing the AI output

Solution

An AI that processes hundreds of applications simultaneously

5 weighted criteria, auto-extracted from the job description

The AI identifies which requirements matter most from the posting and assigns weights. Recruiters can adjust those weights in real time, which is also the main trust mechanism.

Color-coded scoring for instant visual triage

≥70% green, 50-60% orange, ≤49% red. Recruiters read the pipeline without opening a single CV.

Penalty formula for unknown criteria

User research showed recruiters want 'not found' to count against a candidate. The more unknowns, the heavier the penalty: a complete 5/5 analysis always outscores an incomplete one at the same criterion average.

Launching an AI analysis

In one action, the AI reads the job description, extracts 5 weighted criteria, and scores the entire applicant pool. The recruiter sets the brief; the AI handles the rest.

Criteria extraction and pool scoring in one step

Scanning the pipeline

Color-coded match scores turn 200 applications into a scannable list. Recruiters see who fits at a glance, without opening a single CV.

Pipeline view, color-coded triage

The full reasoning on hover

Hovering a score surfaces the criterion breakdown and the radar chart. Enough detail to decide, without reopening the CV.

Score hover: criterion breakdown and radar

Impact

What changed after launch

The scoring system and the interface are designed. The next design challenge is bulk decisions at scale: how do you let a recruiter act on 200 AI-scored candidates in minutes, not hours?

Reflection

What I learned

What worked

Reframing the problem from "improve ranking accuracy" to "reduce cognitive load" was the pivotal decision. It unlocked every design choice that followed: progressive disclosure, adjustable weights, deal-breaker-first hierarchy. Without that shift, we'd have been optimizing an algorithm that recruiters were already ignoring.

What I'd do differently

I would have tested trust in the AI earlier in the process. The opacity problem surfaced during prototype testing, not during research. If I had framed the question differently in interviews — asking 'would you need to verify an AI decision, and why?' — reasoning transparency might have been central to the first design, not discovered as a gap in iteration.

What's next

Two features are in the backlog. A global statistics board will surface adoption and scoring trends across all active job postings, giving HR managers visibility into how the AI is being used at scale. A bulk decision board will let recruiters mass pre-qualify or reject candidates directly from scores after an analysis, addressing the one step that still requires a manual action per candidate even when the AI has already done the work.

Next project

quarksUp DS

Design system audit to rollout at a live B2B SaaS product

Full case study

Protected content. Enter the access key to continue.

Request access via

LinkedInEmail

Introducing AI where time is lost the most

An AI that processes hundreds of applications simultaneously

5 weighted criteria, auto-extracted from the job description

The AI identifies which requirements matter most from the posting and assigns weights. Recruiters can adjust those weights in real time, which is also the main trust mechanism.

Color-coded scoring for instant visual triage

≥70% green, 50-60% orange, ≤49% red. Recruiters read the pipeline without opening a single CV.

Penalty formula for unknown criteria

Launching an AI analysis

In one action, the AI reads the job description, extracts 5 weighted criteria, and scores the entire applicant pool. The recruiter sets the brief; the AI handles the rest.

Criteria extraction and pool scoring in one step

Scanning the pipeline

Color-coded match scores turn 200 applications into a scannable list. Recruiters see who fits at a glance, without opening a single CV.

Pipeline view, color-coded triage

The full reasoning on hover

Hovering a score surfaces the criterion breakdown and the radar chart. Enough detail to decide, without reopening the CV.

Score hover: criterion breakdown and radar

Accelerating recruitment with AIAccelerating recruitment with AI

Introducing AI where time is lost the most

Full case study

Accelerating recruitment with AIAccelerating recruitment with AI

Introducing AI where time is lost the most

Full case study

Accelerating recruitment with AI

Accelerating recruitment with AI