Alex Safonov

Hi, I'm Alex.
I think in systems — and build them.

Engineering organisations, AI platforms, and the operating system I run my own life on. My background is in systems engineering, and one idea from it keeps being true: optimising one part rarely optimises the whole.

Over twenty years that's meant growing Kaspersky's B2B engineering to 400+ people, leading Visualfabriq as CTO and taking Smartcat's engineering AI-first. Now I'm building Peoplos, a new-generation AI HR startup, and working with a handful of founders, investors and leaders.

I live in the Netherlands.

Systems I build

Same habit at three scales: find what's actually slow, then redesign the system around it — not the people in it.

Organisations

Past a few hundred engineers at Kaspersky, I stopped managing people one by one and started designing the organisation — teams, incentives, feedback loops — so the right behaviour became the default. When Smartcat went AI-first, we did it the same way: by changing the structure, not by running a training programme.

Products

Platforms that carry AI into real work: revenue management for Nestlé, Heineken and Unilever at Visualfabriq; LLM copilots and human-in-the-loop workflows in Smartcat's core product, with delivery down from months to weeks. And now Peoplos — HR software rebuilt for the age of AI.

Myself — Personal OS

I built my own Personal OS with AI, to run my life on autopilot. It catches everything that comes in, sorts it, plans my day and tells me what actually needs me. The one thing it never does alone is commit me: no email, meeting or payment goes out without my yes.

  1. Capture agent
  2. Clarify agent
  3. Plan agent
  4. Decide me
  5. Do both
“My bottleneck was never typing. It was decision latency.”
  • I just talk to it — ideas and tasks go in by voice, in English or Russian.
  • Every morning it fits what matters into the free hours in my calendar.
  • Decisions arrive ready to make: options, a recommendation and a deadline.
  • Each week it asks for three wins that matter — and keeps me honest about them.

It's also my lab: whatever works here, I bring to teams.

Autonomy without boundaries is a demo. Autonomy with explicit gates is an operating model.

How I got here

  1. 2001

    Minsk. Java developer and test lead at Kyriba, load-testing a treasury platform — learning that systems break at the seams.

  2. 2003

    Luxoft. Seven years of projects for banks, energy and media, and a development centre for a retail bank built from scratch.

  3. 2010

    Kaspersky. B2B engineering from ~250 to 400+ people, 20+ enterprise products, a virtualization-security line from zero to $10M a year.

  4. 2021

    Visualfabriq, the Netherlands. CTO. The team and the enterprise customer base doubled, and I took the company through technical due diligence for PSG's investment.

  5. 2024

    Smartcat. Took engineering AI-first: platform rebuilt for AI, copilots in the core product, delivery from months to weeks.

  6. Now

    Building Peoplos, on the Board of Advisors at MoonTech, coaching leaders as a Heroic Coach.

Off the clock

I'm usually wet, cold or muddy — by choice. I run Spartan races, play ice hockey, sail and freedive.

Alex with a Spartan Beast finisher medal Alex in an ice hockey jersey holding a helmet and stick Alex at the helm of a sailing yacht Alex in a freediving wetsuit and weight belt

Working together

When an organisation, a platform or your own week keeps fighting you, it's usually a system problem. I help founders, boards and investors as a fractional CTO, an advisor, or with technical due diligence before an investment — and leaders as a mentor or a Heroic Coach.

The first conversation is just a conversation.