Viktor Bezdek

Can you tell us a bit about yourself and your journey at Groupon?

I joined Groupon to make what customers see and touch actually work well—fast, modern, and clean. I run consumer engineering for web and mobile across multiple regions, and I hate when slow delivery gets blamed on "process” and when “process” slows down delivery. I've spent my career building products with a systems mindset, and lately I've been pushing hard on AI to speed things up. I'm here to use complexity as an advantage, not let it become an excuse for missing deadlines.

In your role as VP of Software Engineering, what problem are you personally accountable for solving right now - and why does it matter for Groupon’s future?

Groupon's success depends on how fast we can ship something, see if it works, and learn from it while dealing with old systems, different markets, and the reality of keeping things running. If we're slow or too cautious, our competitors will just move faster.

When everything feels important, how do you decide what truly deserves your team’s time and attention?

I ask four questions before committing my team:

• Does it move a number we care about? Revenue, conversion, engagement—something we can measure, not just a good idea.

• How fast can we prove it works? If we can't validate it quickly, we're probably guessing.

• What are we not doing instead? Every yes is a no to something else.

• Does it fix a recurring problem? One-off wins are fine, but eliminating ongoing pain compounds.

If something doesn't pass this filter, it's probably a distraction even if it sounds important.

What’s one example where saying “no” or simplifying the scope led to a better outcome than doing more?

Last year we nearly killed our new app because we were measuring it wrong. For nine months, every metric showed the new app underperforming the old one. We dug into everything - performance, UX flows, push notifications, edge cases, you name it. The data kept saying we'd built something worse. Turns out we were comparing all users on the new app (mostly new signups) against all users on the old app (heavily weighted toward established customers). Of course it looked bad. We were measuring different populations. Once we isolated just new user cohorts and compared those, the picture flipped. The new app was actually performing better. We'd just been looking at it wrong the whole time. Since then, progress has been much smoother.

A lot of real progress comes from unglamorous but essential work. How do you create an environment where teams take pride in that kind of execution?

I don’t have to do much. I work with amazing people that understand what it means to build and maintain a product.

What do you expect from your leaders and engineers when things don’t go as planned?

When something breaks, I want to know fast. Tell me what happened, what you're doing about it, and own the fix. The person who broke it gets to build the guardrail that prevents it next time. That's not punishment it's the best way to actually learn from mistakes. I don't care about blame. What I do care about is when people dodge responsibility or avoid hard work. Those are the only things that actually bother me.

You’ve been pushing a very hands-on, AI-first way of working, including building your own Claude Code setup to bring clarity across data, priorities, and execution. What gap were you trying to close with this approach?

I've been playing with AI since before ChatGPT, and I'm convinced it's going to change what engineering work looks like. With every model release, more stuff becomes automatable. If a machine can do it reliably, we probably shouldn't be spending human time on it. I use AI heavily in my own workflow - it gives me operational visibility into what's happening across teams that I couldn't get otherwise. It helps me be a better manager.
My primary intention with preaching AI-native mindset is that I  actually worry about people on my teams staying employable. The market is moving fast. Engineers who don't adapt, who don't learn new tools, who don't expand beyond their narrow specialty are going to struggle. And HR isn't going to save anyone. By the time they roll out "AI training programs," the landscape will have shifted again.
I'd rather show people what's possible now than watch them get caught flat-footed later.

How has this changed the way your teams identify problems, make decisions, or move faster day to day?

The immediate impact has been on my effectiveness as a manager, allowing me to dedicate less time to administrative tasks and more to meaningful work. My current focus is on scaling this to benefit the whole company, which I anticipate sharing soon.

Looking at recent initiatives like MBNXT Touch internationally and the MBNXT app in North America, what mattered more than the rollout itself in making these efforts successful?

Honestly, rollout is just logistics. Anyone can ship a release. What actually made MBNXT work, both the North America app and the international expansion, was the stuff you don't put on a roadmap slide. Ownership mattered most. Not "the team owns it" kind of ownership, but clear, named accountability per stream. The moment responsibility gets shared, it quietly belongs to no one.
Then there's measurement. This one's harder than it sounds. You're comparing a legacy app against a new one, and those aren't apples to apples. Knowing what to measure, how to set up the comparison, and how to interpret what you're seeing without fooling yourself takes real deliberate thought. Get that wrong and your feedback loops are running on bad data.

Speaking of feedback loops, tightness matters. Measure, learn, adjust. Not quarterly. Fast. The teams that did well treated every release as a question, not a declaration.
Execution discipline was the other thing. Cut scope, ship, validate, repeat. Not because cutting corners is good, but because shipping something real and learning from it beats perfecting something imaginary.

And the last piece is what happens to a team's headspace when things go wrong. Because they will. Reality doesn't care about your plan.

When that happens, my job is to step in front of it. Absorb the noise, the pressure, the blame if it comes, so the team can stay focused on actually fixing the problem instead of covering their backs.

What’s one belief about leadership or execution that you hold strongly - even when it’s uncomfortable?

Feedback loop is what moves the needle. Radical transparency builds trust (even though it’s often not comfortable). Praise and recognition is non-negotiable.

Finally, what’s one thing about how you work or think that most people wouldn’t guess from your title alone?

Probably that I still do a lot of the work myself. Not to micromanage, but because I genuinely can't lead in this space from a distance. Most evenings, after my wife and son go to sleep, I'm building something. An internal tool, an agent workflow, a prototype I can demo to engineers the next day. Not because I have to, but because I think you can't credibly push a team toward AI-native thinking while staying abstract about it yourself. I show up to technical discussions and actually participate. I share what I'm learning. I demo things I built over the weekend.

If someone was visiting CZ for the first time, what is the one city, landmark, or experience you would recommend they see first—and why?

Tisské Stěny in Ústí nad Labem region. Chronicles of Narnia was filmed there and it's a really beautiful piece of nature.

If you could swap lives with any fictional character for a day, who would you choose and why?

I’m happy in my own skin. To be honest I don’t have such a desire.

What's the most adventurous thing you've ever done, and would you do it again?

Hard to pick just one. Either climbing Opera Vertical in Verdon — a 900-meter wall in the south of France — or hitting 300 km/h on a motorcycle. I'd do Opera Vertical again in a heartbeat. The motorcycle thing, probably not. I have a three-year-old now, and motorcycle racing is just genuinely dangerous. Rock climbing gets a bad reputation, but statistically it's safer than skiing — most people don't know that. On a wall, you're in control of the risk. On a track at those speeds, the margin for someone else's mistake is basically zero. So the wall stays on the list. The bike stays in the memory.

If you could host a dinner with any historical figure, who would you invite and what would be the one question you’d want to ask them?

Nikola Tesla, no question. I'd want to ask him about wireless energy transfer — not the textbook version, but what he actually believed he had figured out. He made claims about the significance of certain frequencies that were either decades ahead of their time or completely wrong, and we still don't fully know which. That conversation would either change how I think about energy and physics, or at least tell me what was really going on in his head.

If you could instantly master any skill or talent, what would it be and why?

Playing piano. I can do it very poorly and I admire people that can just sit and play what they feel.

What's the most embarrassing song on your playlist that you secretly love?

Honestly, none.  I don't really do guilty pleasures. My taste is weird enough that I stopped caring what people think about it a while ago. What I will say is that I accidentally found a hard drive I'd completely forgotten about, full of music I'd been working on 10+ years ago. Dug through it and over last year finished what I felt had potential. It was a great way to relax from stress at work.  It's out on vinyl and on every streaming platform — search Astroform, Singularity Ballet.