ALL POSTS

By Marcell Balogh, CTO / Lead Data Scientist & ML Engineer, Dinkmate.ai Everyone building sports tech right now seems focused on the same thing: where did the ball go. Placement, speed, spin, whether it landed in or out. All useful. All, we'd argue, downstream of the thing that actually matters. A ball can travel from point A to point B and land well or land badly, but the ball doesn't decide that. Your body does. The stance you set up in, the angle of your knees, how far you

By Marcell Balogh, CTO / Lead Data Scientist & ML Engineer, Dinkmate.ai There's a comfortable story going around ML circles right now: get enough data, point a solid off-the-shelf model at it, and the problem basically solves itself. Architecture matters less than it used to — data is king, or so the story goes. We believed a version of that too. Turns out the devil is in the details. Without a real, purpose-built architecture, no amount of data saves you — the ball's just to

Pickleball is the fastest-growing sport around the globe, and clubs and leagues are feeling the growing pains. New members show up faster than organizers can figure out where they belong. Drop into any open play session and you'll see the same problem: a beginner who just bought their first paddle gets paired against a former college athlete, the beginner spends the whole game chasing balls they'll never reach, and the stronger player stands around waiting for a real rally. N

By Marcell Balogh, CTO / Lead Data Scientist & ML Engineer, Dinkmate.ai There's a moment every sports tech company eventually runs into: you've built a beautiful feature — shot detection, rally analysis, ball velocity — and then someone asks the obvious question. Okay, but whose shot was that? If you can't answer that, the feature is a party trick. It looks good in a demo and falls apart the moment someone tries to actually use it. A stat without an owner isn't a stat Every p

By Marcell Balogh — CTO, Lead Data Scientist & ML Engineer at Dinkmate.ai At Dinkmate, we live by one rule: build everything in house. Not for the fun of reinventing wheels, but because owning the full stack is the only way to move fast without depending on someone else's dataset, licensing terms, or roadmap. This post covers one piece of that philosophy: how we generate synthetic training data for pickleball court and ball detection, entirely from scratch. The problem with r

By Marcell Balogh, CTO / Lead Data Scientist & ML Engineer, Dinkmate.ai AI is rewriting the rulebook for sports technology. Building a computer vision or ML pipeline that tracks a player, scores a rally, or reads a swing is easier today than it's ever been. Open-source models, pretrained weights, and off-the-shelf frameworks have collapsed months of engineering work into days. But that ease of development hides a trap: teams start believing the model is the product. It isn't.

At Dinkmate, we’ve always believed that ratings should reflect more than just wins and losses. Ratings should capture how you really play...

Pickleball is growing faster than ever, with over 30 million players across the U.S.—yet the majority still don’t have an official...

Sándor’s connection to pickleball runs deeper than business—it’s about building a thriving community. A passionate player himself, Sándor...

Marcell’s journey into pickleball is just beginning, but he’s picking up the game remarkably fast. With his background in interactive...