About WeLabelData
Clean multimodal data is a craft we've practiced since 2013.
We turn messy, multimodal sensor data into training-ready datasets. We were doing the hard part — clean, structured, multimodal labels — long before it was called Physical AI.
The founder
Two decades of turning sensor data into something models can learn from.
WeLabelData was founded in 2022 by Ivan Pasichnyk.
Before WeLabelData, Ivan was Head of R&D at Kyivstar — Ukraine's largest telecom, and the first Ukrainian company to trade on a US exchange (Nasdaq: KYIV, 2025).
From 2013 he built Ecoisme, an energy-efficiency startup that recognised household appliances by how they appear in the electrical grid — capturing each device across its operating modes, decomposing the signal with Fourier analysis, and labelling it as multimodal data. That's the same problem we solve today, just with a different sensor.
Forbes 30 Under 30
Europe · Technology, 2016 — first Ukrainian on the list
CES Innovation Award
Smart Home, 2016 — Ecoisme
EDF Pulse Award €100K
1 of 3 winners from 300+ European startups
Techstars / Virgin Media
Accelerator alum — Ecoisme
One craft, many domains
For over a decade we've done one thing well: take raw, multimodal data — video, computer vision, sensor streams, audio — and make it clean, structured, and trustworthy enough to train on. We've shipped it across telecom, sports analytics, surveillance and energy.
Physical AI and robotics is the newest and hardest frontier we've moved into — the same craft, a harder sensor. It's where we're investing now, but it's one domain in a broader data layer, not the whole of it.
Built with people
Quality always comes first — and we deliver it with a team that has a real reason to care.
WeLabelData partners with EpiProsvita, a Ukrainian nonprofit that supports thousands of families living with epilepsy. Many of the people doing our annotation are mothers raising children with epilepsy — careful, meticulous work that fits real lives.