Experience

Not a CV. The story of how the way I work took shape.

My titles changed a lot: web builder, project manager, product manager, CTO, engineer, founder. The way I work didn't. I get a problem, think about it, talk to people, resolve it, and look for the next one. These chapters are about how that pattern formed, what it cost, and what it's turning into now that AI is part of it.

Chapter 1 · IBM · 8 years

I learned to make things happen

College didn't feel like preparation for life, so near the end of it I started looking around, and IBM took a chance on me. I started as a web builder, which meant copying information into a content management system.

Deploying to staging needed approval. There was an “Ask for approval” button, and we were told not to use it: ask in the group chat instead. One vacation period nobody was around to approve, and out of frustration I clicked the button. Fifteen minutes later the responsible manager messaged me, glad someone was working on it, and approved. The next time I only clicked the button. It got escalated. My manager called me and said I wasn't following the process, but he was happy I kept finding ways to move things forward. Asking for forgiveness instead of permission stuck with me after that.

Then my phone started ringing: people from the next tower, the upper floors, other countries, asking me to build things. My manager pulled a designer, a manager and me together to pitch an idea to top management. I was handed the presentation. After 25 minutes of talking I said, “You know what's better than talking? Showing you,” and ran a live demo. The manager on the line wanted to call the US right away. That became a new team, Top Talent.

A sales rep came to us with a painful process: service contracts were approved by emailing Excel sheets back and forth, and every objection added another two weeks. My team built a web app for it, and two weeks became two hours. The sales rep said they knew it would save time, just not that much.

By the end I ran my own team, as a manager on paper only. I was known as the person who got things done, but I couldn't call the shots. Watson arrived as I was leaving. It was my first look at AI, and it was nowhere near ready.

It led to: wanting more ownership than a big company could give me.

Chapter 2 · SmartSource Partners, Greenlight

I learned structure, then humility

SmartSource Partners was tiny with major clients. I joined as an ambitious manager and started hands-on, building HTML5 banners, with a promise of a team. A team did get built, just not mine. The owner noticed I see a lot of structure, and made me head of project management. I built a project management and sales team.

Then I moved to Prague as the product manager of a product company just starting out. At Greenlight we researched a product that could have been game-changing, with natural language processing that really worked. I left before anything shipped. I came in new to product management and left knowing what the job takes. Humbling.

It led to: the job where I stopped waiting to be handed problems.

Chapter 3 · Creative Dock

I learned to find the next problem

Creative Dock builds startups inside corporations. I joined one of them, an insurtech product that's still running today, as a project manager. About two weeks in, one morning, I was asked to be interim CTO: the CTO we wanted had turned the offer down. I was a patch. A few days later he accepted, but I was never moved off the role. We split the work with no power struggle. He took the stakeholders and the core product. I hired the core product team and automated whatever hurt internally.

The first customer after launch was our CEO, and we handled his purchase by hand in a CSV template. The next day there were five more, and nobody wanted to retype 50 columns from scattered data. I stayed late, automated the report and presented it the next morning. Then the insurance partner's head of security told us that sending this data around in CSVs wasn't acceptable. A week later we had secure, automated delivery on both sides. Then the partner asked for more reports, and I built a reporting suite.

I also built a small display in our office corner: sales today, contracts since launch. People walked by and said, “138 already?” The 100th contract was an event. Nobody asked me to build any of it, and I didn't ask for permission. My last fix there was finding and training my own replacement.

“Peter is a trooper. Whenever a grenade landed in our backyard, he was the first on site, covering it with his own body.”

My CEO, in his referral

It led to: wanting a real CTO job of my own.

Chapter 4 · Fresh Tracks

Leading without building wasn't enough

Fresh Tracks joined my two loves, music and technology. I took over as CTO from a burned-out predecessor, the person closest to my skill set I've ever met. The problems were interesting. But the stack was new to me, I was asked not to build hands-on, and my days were hiring and strategy. It didn't work.

The music side went sour too. Music all day at work, mixing and producing at night, gigs on weekends: the two things I loved blended into one job. They offered to let me pick my own role. I said no. It was COVID, and I wanted to find myself again.

It led to: knowing that next time, I had to build.

Chapter 5 · Cyrkl

I finally became the CTO

Cyrkl is an impact startup in Prague: a marketplace where somebody's waste becomes somebody else's commodity. I came in with two CTO jobs behind me that hadn't worked out, so I didn't come in as CTO. I came in as a founding engineer, to take an external prototype and rebuild it in-house, built to scale. I was promoted to CTO quickly, got equity, and joined the investment rounds, the investor calls and the board. The moment it felt real was an all-hands where I presented the strategy I'd written.

When we got it right, the impact was immediate: materials that would have gone to landfill got a second life. We started to dominate the local market, moved into Europe and aimed at global. I led up to 40 people.

Then came a nomination nobody explained. A call with Deloitte. A call with someone who turned out to be a senior architect at Google, whose questions I answered bluntly. We won the Technology Rocket Ship Award from Google and Deloitte, for scalability. I wasn't at the ceremony. I was busy with problems. Privately, I worried we'd built something too close to state of the art for a 70-person company to keep up.

We scaled too much, too soon, in a market that went up and down. When it went down, cash flow couldn't carry 70 people. I was in Spain with my partner, seven months pregnant, on our last trip before our daughter was born, when the CEO texted about a crisis call. I joined. It took four hours that day and half the next, and we let half the company go. My team wasn't affected. But fixing a broken workflow and deciding who keeps their job are different kinds of responsibility.

It led to: the first thing that ranked above the problem.

Chapter 6 · My daughter

Something finally outranked the problem

For years it went: problem, Peter, fix. When my daughter was born, there was a person who needed me around more than a startup did. After two weeks of paternity leave, work became nine to five. I finished what I had to as fast as I could and went home. Leaving was a conscious decision. I was still a fixer, but fixing wasn't my first priority anymore.

It led to: trying the job everyone assumed I was best at.

Chapter 7 · Oaks Lab

I found out what I don't want to be

Oaks Lab builds products for startup founders, and they're among the best at it. I was hired as a tech lead, but there was no project for me, so I worked as a senior developer. It wasn't my strength, and they let me go. It wasn't a failure. We put real effort into closure on both sides, I still go to their annual party, and I recommend them to anyone. If I ever have to be let go again, I'd want it done like that.

It showed me where the line is. I can't be only the person writing code. I also can't be only the CTO in board meetings. I do best somewhere in between: get a problem, think, talk to people, build, fix, find the next one. I didn't have a name for that yet.

It led to: the tool that changed what one person can build.

Chapter 8 · Burning Man, GPT-3, Claude Code

AI became mine to use

I'd met AI three times before. At IBM, Watson was broken. At Greenlight, NLP worked, but someone else built it. At Burning Man in 2018 and 2019 I met researchers from a young company I'd never heard of, OpenAI. In 2020 or 2021, at a regional Burning Man event, someone told me about GPT-3, and my OpenAI friends hinted they were cooking something. It was ChatGPT, a year later. I got to try a limited version of GPT-3. It was scary how good it was.

The real turn came from a new tool that only ran in the terminal: Claude Code. I topped up the API with $20 and gave it a task from my backlog. It did it for 40 cents: two hours of my struggling, nearly right. Over the next two or three days I burned through the whole $20 and got more done than I thought possible. The world was shifting.

It led to: letting AI find my next job.

Chapter 9 · Buddy Pro

I taught people until they didn't need me

I believed hiring could be person-centric, with AI helping. So I let GPT help me describe who I am and what I'm looking for, and used those $20 of Claude credits to build a tool around that description. It scraped local job boards and picked the roles that fit. One day it flagged Buddy Pro with a very strong signal. The ad was bombastic: a rocket, millions in revenue, A players only. The process was complicated. At the end I talked to their CTO, and he hired me on the spot.

I was hired to write code, and soon realized someone was already much better at it than me: AI. That identity crisis could have gone several ways. It went the best way. The company was pro-AI, but nobody was doing anything with it, so they let me put my money where my mouth was. I enabled AI across the company and taught people until my role wasn't needed. Then I left.

It led to: choosing my own problems.

Chapter 10 · Now

I choose the problems

Teaching runs through my whole career. For years I thought I had stage fright. In the last two years I found that what I love is the preparation: working out what matters, finding the story, and then going on stage understanding it.

Now I've stopped looking for a role. I'm building my own courses and Moshun Earth, an AI-native impact company, with a co-founder. The fixer is still here. AI changes how much one fixer can do: I no longer have to write every line to stay close to the building.

The arc

  1. IBM taught me to get things done.
  2. That made me the person people called when something needed fixing.
  3. That made me responsible for systems, then for people.
  4. I found that leading without building wasn't enough.
  5. I became a CTO who owned the system.
  6. Then I found I didn't want to be only a CTO, or only an engineer.
  7. AI lets me work in the space between the two.