Hi! I am Daryna

I'm a product designer who loves to create useful and intuitive products. My background is in workflow research and UX for complex, data-heavy B2B products. I enjoy working end-to-end: discovery, problem definition, solution design, and implementation.

Currently

RELEX Solutions

Based in

Helsinki, Finland

Scroll

Selected work

2025–2026

0-to-1 Mobile B2B SaaS Field research

Empowering store personnel to manage orders and inventory efficiently

Led UX for MoRe — RELEX's new React Native app replacing legacy solution for store replenishment. End-to-end ownership from discovery through shipped features.

Read case study →

2025–2026

Feature design Workflow

Improved orders accuracy with guided inventory counts

Designed a structured inventory counting workflow that sequences items to count by physical location in the store, tracks progress in real time, and ensures nothing gets left half-finished.

Read case study →

2025–2026

User research Claude Code

Manual orders: speeding up design process with AI as a partner

Designing and prototyping a workflow for store associates to create and submit orders for products that fall outside automated replenishment.

Read case study →

2024–2026

Enterprise SaaS Desktop B2B Collaboration

Other work

A selection of additional projects: floor plan reviews, planogram workflows, and location management tooling designed for category managers and space planners.

View work →

Let's talk

daryna.barsukova@gmail.com

← All work

2025–2026

Empowering store personnel to manage orders and inventory efficiently

In-store ordering product was identified as critical for US market expansion. The existing tool was not scalable, it also lacked in usability and mobile UI. MoRe was a greenfield project where I had a chance to design a mobile app for the real workflows and real devices.

Role

UX Designer focused on workflow research

Team

Lead designer (summer 2025), mobile designer (from October 2025), 2 PMs, PO, 10 engineers

Platform

iOS · Android (React Native)

Users

Store associates · Dept managers · Store managers

Status

Pilot customer enterprise rollout

The problem

No structure, no hierarchy

Orders in a flat list with no filtering or prioritisation. Associates scrolled through hundreds of items to find what needed attention.

Disconnected modules

Ordering, inventory and product data lived in separate dashboards. Checking all three for one item meant navigating to three different places.

Designed for an office, used on a shop floor

Poor WiFi in backrooms, interrupted tasks, devices put down mid-flow. The tool assumed stable conditions that don't exist in stores.

Devices the design had to serve

Infrared barcode scanners are standard on the shop floor. Device-specific capabilities like Zebra scanning required deliberate interactions.

Product design principles

Shaped together with the lead designer — these guide feature design, engineering conversations, and tradeoffs across the entire product.

01 — Minimise time-to-task

Tasks must be intuitive, quick, and easy to learn. Every tap, scroll, or modal should be questioned: "Does this slow the user down?" If yes → simplify.

02 — Physicality as the core of UX

The work happens in the real world — walking the store, carrying boxes, scanning barcodes. UX patterns must respect mobility, time pressure, and floor plan driven workflows.

03 — Designing for data integrity

The product lives and dies by the quality of store-level data. Designs must make it easy to capture accurate data in the moment, with autosaving, realtime updates, and minimal friction.

Major user stories

Item search

As an in-store user, I need to be able to easily find master data and all information regarding a product so I can make informed decisions and take action if needed.

Inventory

As an in-store user, I need to be able to manage inventory quantities throughout the store, so that I have an accurate capture of how much product exists in-house.

Ordering

As an in-store user, I need to be able to edit order proposals so that I can make sure there is enough product on shelves based on recent events and trends.

Shrink

As an in-store user, I need to be able to report shrink of all types so that I have an accurate capture of how much product has been removed from shelves.

Field research

I planned and facilitated a full-day onsite design workshop at the customer's headquarters in the US in October 2025, getting insights directly from store associates and managers, then rapidly iterating designs based on what we observed.

We also ran weekly co-development sessions with the customer team — sharing designs, getting feedback, and aligning on roadmap. Multi-customer requirements from five retail chains were balanced into a single unified design.

Impact

"This is truly a tool built from the stores up."

VP Inventory Management, US pilot customer

Referenceable customer

The pilot customer became a referenceable customer for store operations in the US market. A key factor cited internally was the quality of the co-designed product development.

Scanning as a primary interaction

Designing barcode scanning as the primary entry point — not a secondary feature — meant associates could move through tasks at their natural pace on the floor, without stopping to type.

Designed to be learnable without instruction

Store associates were able to pick up the app without a formal training session. In a high-turnover environment, that matters.

← Previous

Manual orders

Guided inventory counts

← All work

2025–2026

Improved orders accuracy with guided inventory counts

Designed a structured inventory counting workflow that sequences items to count by physical location in the store, tracks progress in real time, and ensures nothing gets left half-finished.

Role

Feature Designer

Users

Store associates · Dept managers

Scope

0-to-1 · Shipped & iterated

The problem

Inventory inaccuracy is a common challenge for retailers, resulting in unforeseen stockouts and lost sales. When on-hand data is wrong, order recommendations are wrong — leading to stockouts, overstocks, and lost sales that are hard to trace back to their source.

Fixing it requires counts to happen regularly and completely. The system identifies which items need counting and surfaces them to the associate as a guided workflow. The associate counts in a defined sequence, and the data flows back to update inventory accuracy.

Design challenge

Build a counting workflow that survives the reality of store environments: interruptions, variable WiFi, staff jumping between tasks, and no time for training.

Design principles

Three constraints shaped every decision in this feature.

1Speed above all

Every interaction should complete in seconds. Counting is high-volume and repetitive — friction compounds fast across hundreds of products.

2Efficiency of movement

Sequence products to minimise physical backtracking. From milk to bread, not milk to bread to milk again.

3Zero training needed

A new associate picks up the device, opens the app, and knows what to do. No manual, no onboarding session required.

Design iteration

Initial direction — Browsable list

The key insight from customer onsite was that location, not product, is how associates mentally navigate the store — so we flipped the structure to sequence by where you physically are.

✅

Capturing stock in cases and units provides flexibility.

✅

Totals do the math for the user.

❓

Inefficient to check 2–4 different places in the store for one single product. Add possibility to capture by location?

❓

How to provide better visibility of what was already counted?

Shipped version — Location-sequenced view

This version was tested with a store manager at the first pilot store.

✅

Navigating sales floor items vs. backroom with segmented control as a way to count wherever.

✅

Add locations manually.

✅

Clear statuses and ability to pick up counting from where it was dropped.

❌

Confirming every single count is cumbersome. Simplify.

❓

How will department managers count?

Iterated version — New confirmation

✅

Summary view with possibility to drill down to department counts.

✅

Single confirmation screen for all counts instead of confirmation per-item.

❓

Should the confirmation screen also be broken down by department?

Impact

"The system itself is intuitive. It led to better balance on hand maintenance overall. Having guided workflows is really going to help our members day to day."

Inventory Operations Manager, US pilot customer

Cognitive load reduced

Associates no longer had to decide what to count next. They can focus their attention on the actual counting in a high-interruption environment.

Fewer missed locations

Location-sequenced counting respected how staff actually move through the store — reducing the effort and discipline required to stay on track.

Managers got visibility they didn't have before

Real-time progress tracking meant managers could see what was counted, what was in progress, and what was left.

Better counts, better orders

More complete counts feed more accurate on-hand data, improving order recommendations.

← Previous

Rebuilding store operations

Manual orders

← All work

2026

Manual orders: speeding up design process with AI as a partner

Most of the orders are automated by RELEX's replenishment engine for stores to review. However, some orders need to be placed manually by store staff. The goal was to understand what store users actually needed and translate that into a designed feature.

Role

End-to-end product design: research synthesis, problem framing, design principles, feature definition, UX flows, and prototyping in code for engineering alignment.

Users

Store associates · Department managers

Scope

Research → Spec → Prototype

Research synthesis

I ran discovery sessions with three customers from pet retail, grocery, supply and convenience. The sessions covered different retail contexts, store types, and device environments. I used Claude chat as research partner.

Core insight

Stores don't think in terms of "automatic vs. manual orders." They think in terms of "what am I ordering today."

Reflection

Before, my synthesis process looked like this: quick notes after each interview, then listening back through recordings to find patterns. It was time-consuming. This time I fed the transcripts to Claude and started asking questions. What quotes support this finding? How strong is the evidence? What contradicts this? The ability to interrogate the data — rather than just review it — made a real difference. It helped me get to outcomes faster.

Venn diagram: Auto replenishment, Pre-books, Manual orders

Design pillars

I distilled the research into three design principles, each framed around a tension the product had to resolve.

Scan anything, learn something

A scan never returns a dead end.

The shop floor is the primary context

Design for store environments full of interruptions.

Confidence before commitment

Validation at natural pause points, not mid-flow interruptions.

Feature set and UX flow

For each pillar I defined a set of concrete features, specifying what the user experiences, where in the journey it occurs, and how it directly reinforces the pillar. Twenty features in total, grounded in the research. Building on that foundation, I designed the end-to-end flow for creating a manual order from item profile and reviewing it in the combined order view.

Reflection

I was largely working on this project alone, so having a thinking partner made a real difference. For both the feature set and the UX flow, I asked Claude to take a first stab at it, then reviewed, pushed back, and iterated. Starting from an imperfect draft was faster than starting from a blank page. A lot of edge cases came out of that process that I might have caught later — or not at all.

Feature set illustration — 3 pillars with features

Scoping the first step

The research surfaced a wide range of use cases, far more than could be addressed in a single development cycle. Rather than designing everything at once, I used the research to identify the most valuable first step for implementation. Scanning is the first thing store employees do on the floor. This led to prioritising scanning workflow: scan an article, land on the item profile, place a manual order. This flow was a clear high-value starting point to release before building the broader flow.

Prototyping with Claude Code

Once the UX flow was defined, I used Claude Code to generate a working prototype of the manual orders feature. Rather than a static mockup, this gave engineering something interactive to explore and react to — surfacing implementation questions earlier than a written spec alone would have.

Reflection

The hardest part was getting set up: configuring the environment, correcting errors, getting the app running on my machine. That took time. Once that was done, generating the prototype from the UX flow was surprisingly easy. What impressed me was the level of detail. Every click worked. Data was manipulable. Interactions I would not have been able to build in Figma were just there in the code. The prototype wasn't polished — but it was real enough to have a conversation.

Outcomes

Shared mental model

Manual and automatic ordering reframed as two types in one order, not two separate entities.

Prioritised first step

Order from item profile identified from research and validated as the right starting point before the full flow was built.

Interactive prototype

Working prototype in a development branch, used in engineering discussions to validate the interaction design early.

Living specification

A Confluence page covering: what manual orders mean in MoRe, typical use cases, feature descriptions, the full UX flow, lifecycle, and open questions for the next phase.

Reflections

The research revealed that manual ordering exists for real reasons, not as a workaround. That meant we couldn't treat manual orders as a temporary solution to be automated away — it had to be a first-class workflow that respected the store knowledge, constraints, and supplier relationships that make manual ordering necessary in the first place.

Designing for speed and simplicity on a small screen, while scoping to what could actually ship first, was the central tension of the work.

AI accelerated research synthesis, but it wasn't a substitute for being present during the interviews. Being there in person meant I could sense context and nuance and that kept the analysis grounded in reality.

← Previous

Guided inventory counts

Other work