I design AI products that fit into real working days.

UX/UI designer at Hiflylabs, working across B2B and B2C. I bridge research, strategy and design to turn complex systems into interfaces people understand and use, from user interviews to shipped components.

See selected work
FocusAI & GenAI product design
PracticeUX research + UI design
EducationMSc Human-Centered Informatics
Selected work · 2025–2026

Case studies

Previews are simplified wireframe recreations. Client screens stay under NDA.

Concept · MobileNews media2026

A new take on a news app

What could the next version of 24.hu, a leading Hungarian news app, look like? We started from readers and the market, and designed a new concept.

Role: UX/UI Designer · research and design concept

Business problem

News apps compete for daily attention, but readers mostly open them only when news breaks. The app needed reasons to come back every day, a better way to serve readers who scan, a home for podcasts and video, and a clear path to subscription.

Project setup

I worked on the research phase (Hungarian market scan, international benchmarks and the app's positioning among Hungarian news apps) and on the design concept. The concept was turned into user stories, and the team built a working Android prototype from them.

Challenges

Keeping what readers already valued (notifications, saved articles, display settings) while reworking navigation and adding new formats. The design system had to come first, so AI-assisted development could assemble proven, reusable components instead of improvising new ones.

Results

A new concept with five-tab navigation, a top 5 with audio summary, TL;DR news cards, daily games with streaks, a podcast and video hub, and an in-app subscription journey. It was delivered as a working Android prototype with 27 screens and presented to the client as inspiration for their app rebuild.

Learnings

AI-assisted development is only as good as its foundations. A solid design system and well-written user stories made the build fast without cutting corners.

AI workflowMarine insurance2026

AI-driven workflows for complex claims

Claims in marine insurance are long and document-heavy. I supported the design and validation of AI-driven workflows that streamline the process, and ran user testing so claims professionals' needs turned into intuitive, actionable experiences.

Role: UX/UI Designer

Business problem

Marine insurance claims are long, complex and document-heavy. A significant part of claims costs went to external lawyers and technical consultants, while claims handlers spent much of their time on repetitive, manual work. With an integrated AI assistant, the insurer aimed to lower legal costs and compensation, increase recoveries, and free handlers for complex, high-value cases. Faster responses and better advice were expected to raise service quality for members, improve the combined ratio and strengthen the insurer's competitive position.

Project setup

I worked as UX/UI designer, supporting the design and validation of AI-driven workflows and working directly with claims professionals through regular user testing. The project ran in parallel streams, so I worked closely with the data, AI and app development teams to keep the user experience consistent across them.

Challenges

A highly specialized domain with expert users and established routines. As claims handlers took on work traditionally done by external lawyers, the AI had to support their judgment and build their confidence without taking control away from them.

Results

AI-driven claims workflows, designed and validated with claims professionals, that turned their needs into clear, actionable steps.

Learnings

AI should support expert work, not replace expert thinking. The design had to keep claims handlers in the decision-maker role, so they kept questioning and verifying what the AI suggested.

AI assistantInsurance2026

An internal AI assistant for insurance employees

An AI-powered assistant built jointly by Alfa and VIG. I started with user interviews to understand employees' needs, pain points and daily workflows, then designed the interface and built components for the assistant's AI-driven interactions.

Role: UX/UI Designer

Business problem

Employees needed internal knowledge (policies, products, processes) in their daily work, but it was spread across many documents and systems. The goal was an AI assistant that makes this knowledge quick to find and use.

Project setup

A joint initiative of Alfa and VIG. I joined as UX/UI designer, running the research phase with employee interviews, then designing the interface and components in Figma together with the product and AI engineering team. I built the design on the client's existing design system, using its components and extending it with new ones for the assistant's AI-specific interactions.

Challenges

Employees in different roles had very different workflows and expectations. Answers had to be verifiable, and AI-specific states like generating, uncertain answers and errors needed their own components.

Results

A research-based assistant interface with chat history that users can organize into favourites, folders and recent chats, a flexible composer, and source validation and AI assessment shown with every answer.

Learnings

Research matters even more for AI products. The interviews shaped what the assistant should do, not just how it looks.

AI · Mobile & wearableConsumer electronicsPoC

A baby monitor that wakes the other parent

A concept for sharing night-time care more fairly. The phone listens for the baby, AI recognizes whimpering or crying, and the parent whose turn it is gets a silent wake-up on their smartwatch. The goal was to involve fathers more in everyday parenting.

Role: UX/UI Designer

Business problem

Night-time baby care often falls on one parent, usually the mother. The client wanted a product that helps couples share it, and gets fathers more involved in night-time care.

Project setup

The project began with a Proof of Concept phase: AI baby-cry detection and how well it filters out other noise, the effective range of a typical Android phone microphone, and the battery and performance load of a background listening service. In parallel I designed the phone app and the smartwatch experience.

Challenges

Waking the right parent without waking the other, which made a silent vibration on the wrist the core interaction. Alerts had to be reliable enough to trust at night, and the design had to respect the limits of microphone range and battery life.

Results

A phone and watch concept covering the full night: one-tap monitoring, a clear cry alert with a single "I'm up" action, and a morning summary that shows how night duty was shared against the couple's own goals.

Learnings

When a product tracks who does more at home, tone matters as much as features. We framed the data as shared goals and progress, not as a score that could turn into blame between partners.

Mobile & adminLottery2025

Scratch ticket management at national scale

For Patria, I designed the desktop admin and mobile/tablet interfaces of a cloud-based scratch lottery system built for the African market. It covered inventory tracking, transaction validation and winning-ticket verification with DataMatrix codes.

Role: UX/UI Designer

Business problem

Scratch lottery operations needed one cloud-based system for inventory tracking, transaction validation and winning-ticket verification, built for the African market.

Project setup

I was the UX/UI designer for two surfaces: a desktop admin interface for back-office teams and mobile/tablet-optimized interfaces for work in the field.

Challenges

Designing for scale from 100k+ to around 1M daily transactions, for several user roles, and for fast, reliable ticket verification with DataMatrix codes. The interfaces also had to follow complex logistics processes, from receiving ticket packages to distributing them to retailers.

Results

Desktop admin and mobile/tablet interfaces that covered inventory, transaction validation and winning-ticket verification, ready to scale with transaction volume.

Learnings

Designing a critical service for users with low digital literacy means keeping it minimal and straightforward. Every screen should be easy to understand at a glance.

GenAIHealthcare2025

A GenAI medical assistant for clinicians

The platform used LLMs and RAG to turn the slow manual review of patient medical histories into an instant, AI-driven process. I designed an intuitive, accessible interface that fit its complex data-structuring capabilities into the demanding daily routines of medical professionals.

Role: UX/UI Designer

Business problem

Clinicians spent a lot of time manually reviewing patients' medical histories before they could make decisions. This took time away from patient care.

Project setup

A GenAI medical assistant built on LLMs and RAG. As UX/UI designer I worked on the interface that brought the platform's data-structuring capabilities to medical professionals.

Challenges

Turning AI-structured data into something a doctor can read in seconds, under time pressure, while keeping the interface accessible and showing where each statement comes from. The platform also had to stay strictly a decision-support tool: the AI could summarize and structure the patient history, but the interface could never steer doctors toward a medical decision.

Results

An intuitive, accessible interface that turned manual history review into an instant, AI-driven process and fit into clinicians' daily routines.

Learnings

Clients often equate “everything on one screen” with a quick overview. I learned to reframe the discussion around what clinicians need to decide, and to bring prototypes early so we could compare options instead of debating them.

What I do

Capabilities

Research and design in one practice. I start from how people really work, then design the interface around it.

AI product design

Assistants and AI workflows people can trust in their daily work.

  • GenAI
  • Conversational UX
  • AI prototyping

UX research

Interviews, user testing and analysis that turn into clear design decisions.

  • Interviews
  • Usability testing
  • Information behavior

UX/UI design

B2B and B2C interfaces across desktop admin, tablet and mobile.

  • Design thinking
  • UX strategy
  • Figma

Design systems

Components, principles and documentation that teams can build on.

  • Components
  • Documentation
  • Miro · Jira · Confluence
How I work

Process

Each phase ends with something the team can see, test or build on.

Phase 1

Research

User interviews and workflow mapping to find real needs and pain points.

→ Insights & problem map
Phase 2

Strategy

Translate insights into principles, priorities and key user flows.

→ Agreed direction
Phase 3

Design

Interfaces and components in Figma, prototyped and tested with users.

→ Tested, detailed UI
Phase 4

Deliver

Handoff and documentation with product and development teams.

→ Build-ready design
Portrait of Hajnalka Mészáros
About

I design AI tools for people whose work depends on getting things right.

I'm Hajnalka, a UX/UI designer specializing in human-centered design and strategic UX. My background spans computer science, web development and human-centered informatics, so I'm comfortable on both sides of the design–engineering line.

Lately most of my work is on AI products: assistants and workflows in insurance, healthcare and enterprise settings, where trust and clarity matter as much as capability.

2025 – nowUX/UI DesignerHiflylabs, Budapest
2022 – 2024MSc Human-Centered InformaticsAalborg University Copenhagen
2021 – 2022PBA Web DevelopmentZealand Academy, Roskilde
2018 – 2021AP Computer ScienceZealand Academy, Roskilde
Contact

Hajnalka Mészáros

UX/UI Designer at Hiflylabs, Budapest

Emailhajnalka.meszaros@hiflylabs.com
LinkedInView profile