
University of Ljubljana — School of Economics and Business
B.Sc. International Business
Ljubljana, Slovenia • 2019–2023

I build AI architecture that keeps humans in the loop. More precisely, I work on interaction architecture, across three layers: how agents get the right information at the right time (retrieval, vectors, the structure of context); how humans and agents work together (transparency, pacing, deliberate coordination); and what gets surfaced back to the user at the right moment.
I focus on understanding what people actually need and finding where AI can help them specifically. A lot of people are lost in the pace AI is moving at right now. I stay close to the work, so I can give real help to real people.
I care about the team, I work hard, I tell the truth, and I try to understand before I optimise. I'm at my best with people who want clarity, momentum, and outcomes.
Hands-on by default. Asking questions early, surfacing problems before they harden. The work lives between people, product, and technology — being comfortable in all three is what makes the difference.
Sport taught me to read people before reading data. Coaching taught me patience — and that the best intervention is usually a good question, not a good answer. Showing up is the whole thing.
Munich, Germany • Jul 2025 – Present
Munich, Germany • Sep 2025 – Present

Munich, Germany • Jan 2025 – Feb 2026
Berlin, Germany • Apr 2024 – Dec 2024
Stockholm, Sweden • Feb 2022 – Feb 2024
Stockholm (remote from Lisbon) • Jan 2021 – Feb 2022
Seoul, South Korea • Mar 2020 – Jan 2021
The largest AI companies keep losing control of their own systems — Amazon, Anthropic, Google, Microsoft, OpenAI. The pattern is the same: AI-generated code deployed without adequate human review, then a post-hoc requirement that senior engineers approve AI-assisted changes. Co-written with Barton Friedland and Jim Highsmith, this piece argues the failures are not technological but architectural, and that aviation, surgery, and nuclear safety solved a version of this problem decades ago by treating human oversight as essential capability rather than expendable overhead.
AI systems optimise what is specified; they do not determine what ought to be specified. If the objective encoded three years ago was incomplete, the system has been compounding that gap at scale — into pricing, hiring, incentives, and capital allocation. The dashboards stay green the entire time.
AI integration often erodes human judgement rather than enhancing it. When systems replace judgement long enough, practitioners lose the capacity to originate decisions independently. The most valuable organisations in forty years will be those that preserved and elevated human authorship, not those that optimised it away.
In an agent-mediated world, execution becomes abundant. What becomes scarce is direction driven by human intent. When agents raise the standard, they erase the old signal that 'good work' once carried—and scarcity migrates upward, from making things to deciding which things deserve to exist.
Modern organisations have systematically moved consequential decision-making away from human judgement, automating critical choices upstream while leaders remain nominally responsible for ratifying predetermined outcomes. The displacement creates the illusion of control while eliminating genuine strategic choice. Co-written with Barton Friedland — formative vs corrective intervention as the lever for redesigning systems that preserve discernment while decisions remain malleable.
Using AI at work is not cheating—when it is used to augment human judgment and capability, not replace it. A framework for organizations to permit, equip, and audit AI-assisted work.
AI isn't lacking—most of the time, it's a you-problem. Bring vague questions and thin knowledge, and AI will upscale your vagueness. This article explores the Augmentation Arc: Mirror, Lens, Lighthouse, and Prism—four modes that transform how we work with AI from reflection to resonance.
Steve Jobs called the computer a 'bicycle for the mind'—but that metaphor no longer fits. In the age of AI, leaders need to think less like cyclists and more like sailors. This piece explores how augmentation requires collective intelligence, turning ambient signals into strategic advantage.

B.Sc. International Business
Ljubljana, Slovenia • 2019–2023

Erasmus Exchange — International Business
Lisbon, Portugal • 2022–2022

Bilateral Exchange — Business Administration
Seoul, South Korea • 2021–2022

Upper Secondary (Gymnasium)
Ljubljana, Slovenia • 2015–2019

Meta Backend Developer
Meta
Backend developer specialization

Project Management Specialization
Google Career Certificates

Programming Using Python
GO TEL
Placeholder until authoritative logo is confirmed

Sales & CRM Overview
Salesforce

Leading Teams: Developing as a Leader
University of Illinois Urbana–Champaign

Introduction to Negotiation: A Strategic Playbook for Becoming a Principled and Persuasive Negotiator
Yale University
Thanks for reading — onward.