Tomasz Czarnecki Resume ↓
AI Product Manager · Pricing · Promotion · Marketing Spend

I make black boxes you can argue with.

Commercial teams don't need the right number.
They need a call they can defend.

Tomasz Czarnecki

First product person at Heineken's Global Analytics & AI Hub. Three production decision systems, built for a ~€2.9B budget.

What the box is thinking

It's all probabilities.

Every model, market and building I've worked on gave a range, not an answer.
My job is getting people to act on the range.

01Intro

Ninety seconds, in person.

What I built at Heineken, what I learned about getting people to trust a model, and what I'm looking for next.

I've been the first person inside something new three times. I was the first hire at a Chinese architecture firm's European office and set it up in Poland from zero. At MAD Architects in Beijing, I was one of two founding members of the parametric design group. And I was the first product person in Heineken's Global Analytics & AI Hub.

02Case studies

Heineken spends €7.9 million on marketing and sales a day.

Today so far · since midnight, your time
€0
€2.9B a year ÷ 365 · FY2024 marketing & selling · Source ↗

Someone decides where every one of those euros goes. Before the hub, that was spreadsheets and gut feel.

03The loop

Three questions. One loop.

01 · Allocation AI
Where?
Wherever the response curve says the next euro pays back best, by OpCo, brand and channel.
02 · Promo Advisor
What?
Whichever offer, mechanic and price simulate the best lift, tested before they ever launch.
03 · Smart Flighting
When?
Whenever the AI response curve says that spend actually converts, week by week.
Measured lift, fed back 04 · next year's plan starts here
One system per question
04Build vs. buy

Three of us against a consultancy's finished product. We won.

Build or buy. Heineken was ready to buy a consultancy's promo-planning product. Three of us got less than half a year to show that building it in-house would work better.

Top consultancy{{ bbThemTag }}
A full engagement team. Partners, consultants and a ready-made product.
Internal hub{{ bbUsTag }}
● Product owner● Data scientist● Me, product
05Deep dive

Buildings that were programs.

Five years working in China, back and forth ever since. I wrote the scripts behind a facade and an opera house interior at MAD, and competition wins at AREP. A decade before "AI agent" was a phrase, I was building evolutionary solvers that bred building geometry from a fitness function.

MAD Architects · Xiamen, 2015–2016

Xinhee Design Center

Skin and bones

A design center for the fashion group Xinhee: six office arms around a central atrium, wrapped in a floating PTFE skin.

My part: I built the parametric model that generated the double-curved membrane facade and made it possible to build. There was no precedent to copy, so the model optimised for two things at once: membrane cost and buildability.

Live · Galapagos evolutionary solver

Watch the facade evolve

Floors stay fixed. The facade is bred generation by generation, wild at first, then converging on the shape that got built.

Drag to orbit · Scroll to zoom

Optimal was easy. Buildable was the job.

MAD Architects · Harbin2015

Harbin Opera House

My part: parametric timber interior, detailed down to fabrication.

Harbin Opera House from above, curving along the Songhua River wetlands
AREP · Shenzhen2016–2017

Gangxia North hub

My part: parametric skylights and entrances for the competition win.

Gangxia North hub: the 'Eye of Shenzhen' lattice skylight above the main concourse
If this resonates, let's chat

Got a black box nobody uses?

Tell me what it predicts and who ignores it. I'll tell you what I'd build first.

LinkedIn ↗ Email
Tomasz Czarnecki © {{ year }} · Designed and built by me with Claude Code. · czarnecki.ai