01
Budapest · 2022 to today · Sales Director Eastern Europe
Growing a channel that had stopped growing
Sales
Leadership
AI I built
ResultA channel that had not grown for many years is growing above the market for the second year running, with dealer satisfaction at its highest in three years.
I changed how our management team worked before I led it: shared Kanban boards, Miro workshops, Mentimeter on every call. The team said it became more efficient, and more fun. The highest performance comes when people enjoy working together.
In October 2023 I took over a channel that had not grown for many years. Before changing anything, I asked everyone in it, and then every dealer in the region.
What the study was worth
~600dealers asked, in ten countries and ten languages
~300answers: half of all dealers, thanks to the follow-up built into the flow
1 weekfrom survey close to findings, instead of about six by hand or through an agency
~155 hof manual work replaced per round: two people for two to three weeks
~200people in the regional organisation received the findings
4 / 5field-force support for the CRM roll-out I lead across 10 countries
Few conclusions, but strategic ones, and they have kept their weight. The study settled how we differentiate the channels, how we price for each, and which strengths to build with our partners. It also overturned a long-held assumption: a channel believed to sell mostly online turned out to sell mostly offline. The direction became clear enough to win buy-in for everything that followed.
What dealers told us became the new channel strategy, designed together with volunteers from my team and from our counterparts, and then implemented. The channel is growing above the market for the second year running, and dealer satisfaction is at its highest in three years.
Adoption is the other half of the work. The CRM roll-out across 10 countries was designed around what adds value for the salespeople, and the field force rated its support for the roll-out 4 out of 5.
What came next
- A commercial offer pipeline, which I also built, taken from idea to analysis and a live test. It finds the best sellers in a dealer's country that the dealer is not yet buying, applies the marketing policy, and checks the live promotion plan to avoid double bonuses. A person reviews every offer before it goes out. It is built to scale to the whole sales force of about 40; for now we are still tuning what converts best.
- AI ambassador for my organisation, moving routine work to automation.
- Relaunching terms and conditions across Eastern Europe (discounts, rebates, payment terms and expected contribution) as part of a global redesign I contribute to.
- Novelties Distribution: every salesperson tracks new product launches from day one. Time to market fell from 6–9 months to 1–2, and new products now reach around 80% of distribution, against 20–30% before.
- A selected dealer network: the target was 200 dealers; more than 300 signed, and the shelf-space target was exceeded too.
- Sales forecasting, as in every leadership role I have held.
02
Shanghai · 2018–2022 · Head of Aftersales Emerging Markets
Redesigning one of the most complex departments
Transformation
Leadership
Customer
ResultAn NPS of 75, an exceptional level for a service organisation, reached in 1.5 years instead of the planned three, after redesigning a department of more than 200 people region by region.
More than 200 people on four continents: emerging Asia (China, India and ASEAN), Eastern Europe, Africa and Latin America. Every other department had been transformed already. Aftersales was postponed: complex, connected across the whole world, and understood by very few people.
Instead of drawing a new org chart, I started with everyone who depended on us. With an internal management consultant, we spent many sessions with stakeholders at head office: what did they expect from Aftersales, now and in future? Then we piloted: Greater China went through an organisation-design framework in a series of workshops, and we found the tasks ahead and the roles we were missing. Leadership's feedback was very positive. We moved on region by region, adapting the framework each time and building a change plan for each: reassigned roles, new ways of working.
Along the way I completed a three-month action-learning programme in organisation development, building a real business case with a mentor.
I visited every region before deciding anything. One harmonised model would never work. Each region did something better than the rest.
Agreement was easy. The hard part was the budget: reinvesting profit today for loyalty that pays back later.
So what moved the score was not money. It was simplicity, and following up on every unhappy customer.
Then I prepared for what was coming. Longer-lasting, more affordable products need fewer repairs. To keep growing, we had to offer users something on top. So we built an innovation team on design thinking.
03
1996 · and again since 2024
Building again
Building · AI
ResultAn independent AI engineering project, designed, built and quality-checked end to end on my own. What I learn there, I bring to work as fast pilots.
1996
A pension fund's payout module, written as a student. I learned the undocumented codebase from the code itself, and programming from books. It paid real pensions to real people.
* FoxPro, 1996
monthly = balance / months_left
IF term = "LIFE"
months_left = life_table(age)
ENDIF
Illustrative snippet
Today
An independent AI engineering project, designed and run end to end on my own.
- Models from three providers, chosen per task on the evidence of a documented comparison
- Answers grounded in a curated knowledge base, not in what a model happens to remember
- 76 automated workflows, generated from code rather than edited by hand
- Quality assurance of AI output before every release
- Data erasure by design: a deletion reaches all the way into the knowledge base
- 62 written design decisions; runs on web, Telegram and iPhone
For the technical reader
- Multi-provider model routing: Google, Anthropic and OpenAI, chosen per task on the evidence of a documented comparison.
- Retrieval-augmented generation over a vector database with seven collections.
- 76 workflows generated from 23 Python scripts.
- Prompts as versioned infrastructure: kept as files, synced to a registry, checked before every push.
- LLM-as-judge evaluation: five synthetic personas walk every changed flow, and a separate judge model scores the transcripts against a fixed rubric.
- 62 architecture decision records; GDPR erasure that reaches into the vector store.
- Web, Telegram and an iOS build running on device through TestFlight, built in the cloud from Windows, without a Mac.
04
Moscow · 2014–2015 · Head of Marketing, pricing owner
Pricing through a currency crash
Sales
Transformation
A system before AI
ResultThrough a currency crash, prices changed almost every second month, and market-share growth beat the market by 2.5 percentage points without harming profitability.
In December 2014 the rouble collapsed. The decisions fell to our team on the day itself: block deliveries, stop every promotion, start adapting prices, and tell our dealers openly how we would work from now on.
In chaos, the most valuable thing we could offer our partners was to stay predictable.
Fast pricing had been impossible. Many product groups against many competitors, fixed logic between versions, prices linked across a whole regional hub, separate lists per channel. So we built the logic first.
We changed prices almost every second month, which had not been possible before, and outperformed the market by 2.5 percentage points of market-share growth without harming profitability.
How I lead
What my team says
From my most recent 360° feedback. I lead 22 direct reports today; in Shanghai, more than 200 people across four regions.
I keep my commitments.The one item every team member who answered scored at the top.
I talk to everyone, at every level.Several people said it independently in their written comments.
I execute.Decisive delivery against targets, rated just as high by every group.
My team rated me higher than any other group did.On every perspective. Direct reports are the hardest audience in a 360.
LanguagesEnglish fluent · Russian native · German conversational
EducationSpecialist Diploma (MSc-equivalent) with honours, Information Systems in Economy
Leadership trainingLeadership development for young executives (2007–08) · Organisation development, action learning (2022) · Design thinking, Hasso Plattner Institute (2016)