Skip to content

Senior Product Designer · 8+ years designing B2B SaaS and enterprise products across industries. From research to shipped software, now built with AI agents.

Eight years designing products people rely on, from property management and healthcare to payments, giving and enterprise consulting. I find the real problem, get product, engineering and the business behind one direction, and design until the hard part disappears.

Domains I've worked in

  • Property managementOperations software for hosts and property managers
  • HealthcareClinical operations and patient-facing workflows
  • Payments & givingDonations and payments at consumer scale
  • TransportationFleet and passenger operations
  • Enterprise consultingAdvisory platforms for global consulting firms
  • Fintech & accountingAccounting, reconciliation and money movement

What I do, in one pictureComplex in.
Simple out.

Before · when an app's data goes wrong

3 teams · 4 days · 340 fixes by hand

Redesigning

Designing for when it breaks

After · the problem explains itself

4 steps · 1 person · no waiting

Numbers look wrongAn alert says what broke
Customer emails support
Support asks engineers
Wait three days
Someone digs for the causeSee everything affected
Download a spreadsheet
Fix 340 rows by handFix it all in one go
Upload it againCheck, then confirm
Now there are duplicates
Clean up the duplicates

Illustrative example · scroll to simplify it

Before: when an app's data went wrong, a customer noticed, emailed support, support asked engineers, everyone waited three days, someone dug for the cause, downloaded a spreadsheet, fixed 340 rows by hand, uploaded it again, and then cleaned up the duplicates. After: four steps one person finishes on one screen: an alert says what broke, you see everything affected, fix it all in one go, and check before you confirm.

Data and research find the problem, workshops align the team, design and AI build the fix, code ships it.

  1. Product data

    Who the users are, and where they drop off.

    • Pendo
    • Google Analytics
    • Omni

  2. User research

    Why they drop off, in their own words.

    • Useberry
    • Google Meet
    • Loom

  3. Workshops

    Teams agree on the problem, then track the work.

    • Miro
    • FigJam
    • Jira
    • Confluence

  4. Design

    Directions, systems and proven patterns.

    • Figma
    • Figma Make
    • Storybook
    • Mobbin
    • Awwwards

  5. AI build

    Agents write, review and generate. I direct them.

    • Claude Code
    • VS Code
    • Cursor
    • Multica
    • ChatGPT
    • Midjourney
    • Nano Banana

  6. Ship

    Real code, deployed and live.

    • GitHub
    • Vercel
    • Supabase
    • Cloudflare

Data shows where it hurts. People explain why. I design the answer, test it and see it through to what ships. Here's one problem, start to finish.

  1. 01 · Signal

    The data says something's off.

    I start in the product data. Plenty of teams sign up. Few ever reach the moment the product is actually useful.

    • Pendo + Google Analytics + Omni
    • Claude Code

    AI's part: An agent builds the funnel. I decide what it means.

    Funnel · new teams · 30 days
    100%
    Sign up
    81%
    Workspace
    47%
    Connect data
    19%
    First result

    −34%

    teams give up here

    Found it: “Connect data” loses a third of new teams

  2. 02 · Listen

    Numbers say where. People say why.

    I interview the teams who gave up and watch their sessions.

    • Google Meet + Loom
    • Claude Code

    AI's part: Agents handle recruiting and transcripts. The conversations and conclusions are mine.

    User interviews
    Synthesis · tagged by agent

    Confusion

    Which integration?

    Too many fields

    Waiting

    Waited days for IT

    Back to spreadsheets

    No value yet

    Never saw the value

    Nothing to show my team

    They quit before they see any value.

    Insight

  3. 03 · Define

    One problem. One metric.

    I write a one-page brief: who it's for, what success looks like, what's out of scope. Product and engineering sign it.

    • FigJam + Miro
    • Jira + Confluence

    AI's part: An agent does the first draft and checks it against the codebase. I shape it from there.

    brief.md · drafting

    Value before setup

    Problem

    For

    Out of scope

    Checked against the codebase: feasible

    Success metric

    2×

    teams reaching a first result in week one

    PMEMYC Signed

  4. 04 · Explore

    Three directions. One choice.

    I design three real directions, in code and in Figma, drawing on proven patterns from Mobbin. Then I choose one and say why.

    • Figma + Figma Make
    • Mobbin + Awwwards
    • Midjourney + Nano Banana

    AI's part: Agents build the working versions and critique them before my review.

    Patterns
    3 directions · built in code

    Sample workspace

    Checklist

    Guided call

    Critic: B and C don't fix the wait for data

  5. 05 · Build

    Built in real code.

    I take the chosen direction into real components in the real repo. I plan it, review every screen and hold it to the bar.

    • Claude Code + VS Code + Cursor
    • Multica
    • Storybook + GitHub

    AI's part: Planner, builder and reviewer agents do the routine work, each running a skill I wrote.

    design/sample-workspace

    1<Workspace sample>

    2 <Header />

    3 <TemplatePicker />

    4 <EmptyState />

    5 <InviteBar />

    6</Workspace>

    Reviewer: banner contrast fails AA
    Builder: fixed and re-checked
    Live preview

  6. 06 · Test

    Real users. Real prototype.

    I put the working prototype in front of real people, not a mock-up. Then I make the call: go, revise or stop.

    • Useberry
    • Claude Code

    AI's part: Agents run the test and summarise what people did and said.

    Prototype test · live
    Tasks

    • Get to a first resultMost

    • Invite a colleagueAll

    GoReviseStop

  7. 07 · Ship & measure

    Ship it. Then prove it worked.

    I hand engineers a reviewed pull request. We roll out behind a flag, and I watch the same funnel we started with.

    • Vercel + Supabase
    • Pendo

    AI's part: An agent watches the metric and flags anything unusual.

    Pull requestMerged
    →
    10%50%100%
    Teams reaching a first result · week one
    Release

    2.1×

    Weekly digest sent: activation up, setup tickets down

The full playbook: skills, agents, process and handoff.

Read the full playbook

A few products I've shaped. Full case studies are on their way.

Swipe for more

I design for the moment someone stops noticing the interface and starts trusting it. Agents help me build faster. The care for the person on the other side is mine.

Portrait of Yelyzaveta (Lisa) Chyhryna, arms crossed, in a dark violet suit.

Hi, I'm Lisa, a Ukrainian product designer and builder based in Belgium. I turn messy, high-stakes problems into products people trust, from first interview to production code.

years in product design
workshops facilitated
teams on my design system
projects built on it
  • People and data
  • Systems, not screens
  • Prototypes in code
  • Decisions made easy
Learn more about me
Download CVsoon