45
specialists from various design disciplines
My main focus is to transform design from a chaotic service into a transparent business tool through design systems. I build my work on trust and clear metrics so that the team can grow, and the products deliver real value to the business and users. I am drawn to systems thinking and practical work, to design systems and processes where everything is interconnected and the product and brand reinforce each other.
I have worked as a designer since 2009 and have held leadership positions since 2015. For eight years, I led design at AUTODOC, the largest online retailer of auto parts in Europe, a company of €1.8B in revenue across 27 markets. I joined a company where design did not exist as a separate function: designers were scattered across different departments with no structure, and I built a department out of that.
How I lead. I believe in direct conversations and open communication, inside the team and outside it. We are all doing one big thing together: helping the business grow, and growing as specialists along with it. Whether it is an idea or a problem, I am always ready to talk it through and help. My references will tell you more than I can.
What I work with. Building and scaling design organizations, turning ad-hoc teams into structured departments, DesignOps, mentoring, design strategy and stakeholder work. Design systems: component libraries, tokens, guidelines and governance, architectural layers. Product across web, mobile, CRM and back-office, B2B platforms, with UX systematized and accessibility built in. Brand: identity, brand books, icon and illustration libraries, motion, packaging, multilingual content. And AI in practice, from Figma plugin development and prompt architecture to image and video pipelines and brand characters.
Tools. Figma, FigJam, Adobe Creative Suite, Weavy, Nano Banana / Gemini, Kling, Seedance, Synthesia, Jira, HTML/CSS.
Out of chaos come systems, teams and products. From building design organizations and brand systems to product redesigns and custom plugins.
When I arrived, “design” was just a few people, each living in their own design world, like a little prince on his own planet. There was no unified tool. There was no clear flow for any given task. There were constant backlogs and deadlines that had long passed. No one was happy with it: designers, managers, or the business. We needed a clear flow to move forward at an even pace.
We started by settling on one tool that covered 80% of our needs. Next came the assessment of each employee’s skills, which allowed us to define our focus areas. For the product: B2C, B2B, skins, CRM/back-office, corporate, private labels. For the brand: promotions, motion graphics, POS, retouching, iconography and illustration, packaging. For each focus area, we prepared our own flow, guidelines, rules, and process for delivering results. On top of this, I grew the leadership layer from two to about eleven and promoted two senior managers from within. I introduced skill matrices, grades and career tracks, then DesignOps: task intake, syncs, onboarding and a single source of truth for assets.
A department that partner teams trust: 87% rated the quality of design as high. Most employees received promotions; the leadership pool grew from within rather than being hired externally. The quality and consistency of the work delivered in line with the brand increased by 37%. The volume of tasks completed before deadlines rose by 68%. In 2025, the motion team produced over 20k videos, which is more than in all previous years combined.
At the start, the entire brand relied on a minimum: a logo, a few colors and a font with problematic line spacing. For a company of this scale, that was insufficient. The atomic groups (colors, fonts, grids, icons, illustrations, photos) existed separately, without a common language, without a single source of truth, and each channel reinvented the visual identity. The team built the system layer by layer, from the bottom up.
Every brand needs a foundation: what it is based on, what it communicates to consumers, both internal and external, what it is associated with, and what its advantages are. A visual anchor is also necessary, one that appeals to the consumer and is convenient to work with. We redesigned the logo. We defined the colors, their usage, and codes for different types of media. We selected fonts that support all necessary languages and have a clear, comprehensible design.
The brand’s vocabulary should be extensive, as it demonstrates the brand’s value. Icons are a clear visual language that communicates simple visual messages to the user. The idea of atomic layering emerged while building this library: a simple icon transforms into a branded one, becoming part of an illustration, and can also exist in a 3D variation. One atom grows into a system and gives rise to the next.
At some point, we all moved the same line one pixel to the left and right. This is a problem of visual perception and a sense of fullness, and of the time a designer could have spent on stronger visualization. For each channel and product, we created a grid system from which we build margins and layouts.
From this foundation, a system of guidelines emerged for all channels: YouTube, TikTok, Instagram, promotional banners, ads, emails, POS, presentations, packaging. Each channel has its own voice, but the brand is always recognizable. The system has continued to grow like a living organism: additional colors to support the main ones, a library of images in its own style, branded auto parts to distinguish our banners from those of partners. We followed the same path for RIDEX and goCORE.
From product to product, the basic components and the logic of working with them varied. There was no consistency with the brand. Developers would recreate their styles for each product, wasting time in the process. Then another designer would choose a different color, a manager would decide to increase the font size or change the icon, and this led to chaos. We had the notes, but we needed music.
It was not easy. The design team began to assemble a unified component library for web and app resources: the website, application, PRO (B2B), Club, corporate website, CRM systems, internal services. We took basic components from each product and brought them together. Each component was created manually, establishing the visuals and logic of operation, and compiling style collections. We wrote guidelines for developers, checked the implementation, added necessary attributes to the components, and so on, iteration after iteration. The components, along with the styles, provided the right tone and a single source of truth. The library was additionally brought in line with WCAG 2.2 (AA) accessibility standards.
As we expanded the library’s reach across projects, we decided to enhance it with skin styles. Skins are copies of the main website, launched to increase reach and sales. Adding skin styles helped developers quickly adapt skins to be responsive and saved time on writing components from scratch. Each skin differs from AUTODOC while still being based on a common foundation: a single set of bricks from which different facades are built.
A single source of truth has allowed us to redesign the AUTODOC website and launch new services in a more compressed timeframe. The company is now moving toward an ecosystem product, and for that a single component library is not enough. An architectural layer is needed. It defines the surfaces of the product and how they behave when used, establishing a unified logic for both web and application. With its help, we conducted an in-depth UX analysis of user interaction with the products. This is an artifact for the future transition. I completed the groundwork, but the rollout fell outside my time at the company.
Design is no longer a bottleneck. With a single source of truth, the team has revamped the website and accelerated the launch of new services; products and projects now appear consistent and aligned with the brand. Developers have stopped having to reassemble components for each project, and the misalignment between design and development has decreased by about a third. The architectural layer will assist in transforming into an ecosystem product.
Product design at AUTODOC was not approached as separate tasks, but rather by domains within the main ATD skin, web and mobile. Under each domain, the product designer managed their area alongside the product manager. I would get involved in blockers and contentious decisions, helped refine the interfaces, and ensured everything adhered to a single system.
The largest share of work went into search, where almost every purchase of auto parts begins. Redesign of the product detail page (PDP): restructured information hierarchy, key characteristics of the part highlighted, and a dynamic price calculation for the set. Product listing page (PLP) and catalogues: simplified navigation, visual autocomplete with brand and category thumbnails, sections for related products and chemicals, and separate branches for tyres, rims, and motorcycles. Custom auto-selector: car selection with prompts, synonyms, and popular models by country.
The other domains: cart, checkout and promotions, with a simplified checkout, free delivery progress, a unified pop-up service and clear cookie consent management; payments and prices, with local payment methods by country and dual pricing for regional regulations; post-sales, with the returns portal and status tracking.
Beyond ATD, the same system maintained versions of the storefront (skins), from the German PKW to regional ones with dual pricing, promotional regulations, and accessibility requirements. Also the B2B platform PRO and the content platform Club, as well as internal CRM systems.
The three stories below may not be the loudest, but they showcase diversity: a B2B platform built from scratch, a unified cart, and a returns portal.
One of AUTODOC’s strategic goals was to enter the B2B market, and the resources were available. However, B2C solutions do not suit the business: workshops order frequently, in bulk, and on a regular basis.
We developed PRO from scratch: a web platform, a mobile CRM for field agents, and a back-office CRM. In UX: selecting spare parts for multiple vehicles at once, searching by VIN, OEM, and registration numbers, and returns in 4 clicks.
Figures from flows still in progress: registration 43.6% → 50.5%, order processing ~39% faster, with 80% satisfaction with the dashboard. The direction grew by over 110% in 2025.
The process was a multi-step wizard (5 steps). Each “Next” was a reload and a risk of abandonment. The cart was cumbersome.
Solution: the team simplified the path, systematically eliminating steps by combining and removing unnecessary elements, everything that distracts the user from the key action: purchase. This is currently in the implementation stage. The next step is a single-page cart: everything needed in one place, without transitions.
Result: record YoY growth in conversion to payment across all three platforms: +6.46 p.p. in Q1, +4.32 p.p. in Q2.
Previously, returns were a static form: manual entry of orders and items, reason in text, scans, waiting for an email. The status was a “black box”. Finding the flow was difficult, launching it was unclear.
Solution: design and UX of a unified self-service portal: a pre-filled list of items with checkboxes, tree selection of reasons, auto-labeling, tracking statuses (Open → In Progress → Completed).
Result: support requests dropped by 32%, post-purchase CSAT increased by 14%. Works in almost all countries.
Each plugin addressed a specific bottleneck for the team and management. I developed them as a discipline, with cost analysis, an implementation guide, and a trial period, rather than based on hype.
The team regularly encountered bottlenecks where there simply wasn’t a ready-made tool, or it was too expensive for the volume we were handling. Designers were losing hours on manual routines: inserting translations, resizing, preparing concepts, and searching for references. This time could have been spent on the actual work rather than its maintenance.
Instead of enduring the routine or waiting for a budget for third-party services, I built the tools myself: plugins that I then integrated into Figma, so they existed in the same workflow where the team operates. Each was born from a specific task and went through a benefit calculation, an implementation guide, and testing. This also helped to smoothly introduce AI generation into our workflow, since the team learned it through familiar tools rather than starting from scratch. Here are three of them.
Both managers and designers needed to quickly create concepts and final designs for seasonal promotions, holidays, or simply category banners. It’s nearly impossible to write a good prompt from scratch: you have to consider the camera, angle, and lighting of the scene. Finding references is also a time-consuming task. But what if you just want to experiment with a direction? The Prompt Builder plugin: you select from ready-made presets for location, lighting, camera, style, and necessary props, and that’s it. It compiles a ready prompt in both text and JSON format, which you can use in any AI image generator. Concepts are created easily, leading to a deeper understanding of the scene and the outcome. A library of prompts with an interface inside Figma.
After the banner concept, the strings were manually sent for translation to another department via spreadsheets, a monotonous task. Crowdin eased the process, but the insertion remained lengthy and manual: copying each banner, selecting the necessary string and the required translation. Solution: the plugin sends and receives translations and automatically creates copies of the creative, names them, and inserts the translation. Preparing the banner in 22 languages × 4 resizes was reduced from ~4 hours to ~2.5 hours, and applying translation changes across 22 languages was cut from ~1.5 hours to ~30 minutes.
You have the character’s portrait, their profile and their outfit. To get the shot you still have to write a long prompt in the model, and the result is often not what you expected, so you go back, check the prompt and generate again. Solution: the plugin removes all of this. You take any photograph and replace the person with your signature character. Or you choose a character and describe the scene. Everything is ready in Figma for creative work.
Plugins have removed manual routines and have become the team’s entry point into AI, integrating naturally into daily work as a tool. Currently, some team members are already writing plugins to simplify and automate routine issues. The tool has evolved from a personal initiative into a common practice.
Humanizing the brand is a trend that showcases our human side. We are social beings: it is in our nature to trust people who are like us. However, organizing a shoot involves not just time and money, but a complex system: finding a model, signing a usage contract, hiring a location, creating a scene, setting up lighting, ensuring team support, and processing the results. And what if... anything can go off-plan. Then there’s the wait for processing and editing photos, which can take a long time and definitely disrupt the content schedule.
With AI characters, we build an image based on the brand’s needs; we know their personality, we can describe their appearance, style of clothing, age, weight, etc. The character doesn’t get ill, can be anywhere at any time, and can do whatever you specify in the prompt. The work took about two months: creating each character and their image, defining the audience they are made for, and running trials of generation and content production with the finished characters. For any generation with a character, we use mandatory markers according to Article 50 of the EU AI Act.
The personas have replaced the organization of shoots and the associated routine. For RIDEX and goCORE, a set of recognizable AI heroes has emerged, allowing the team to produce content without models, locations, lighting, and venue budgets. The human face of the brand has become available on demand, rather than on a shooting schedule, and is not dependent on numerous unforeseen factors.
Building working plugins was only the beginning. The same logic can grow into a service that lives inside a company and inside its product, covering a part of content generation: product cards, promo imagery, lifestyle shots, catalog operations with product photography.
The service reads the product and helps show it at its strongest, with brand layering, the right sizes, styles and characteristics. The output goes wherever it is needed: the product page, ads, social, email, design.
What matters here is seeing early where a business loses time, and closing that gap together with the team.
Results feed back into planning: the system sees what worked and adjusts what to generate next.
A concept, not a finished product. There are plenty of stages between the two, with their own ups and downs. The point is something else: to show a view where brand and product are connected by more than a palette and a typeface, where each one makes the other stronger and both work for the business.
Open to Head of Design, Head of Brand and design leadership roles.
Relocation / remote · Romania, Portugal, Ukraine, EU