Background
Polish engineer who co-founded ElevenLabs in 2022 with Mati Staniszewski, reportedly motivated by badly dubbed films. The company went from a $2M pre-seed in Jan 2023 to $500M at an $11B valuation in Feb 2026, shipping voice cloning, AI Dubbing, Projects, Voice Isolator, conversational agents, Scribe, Eleven v3 and Eleven Music along the way.
ElevenLabs is the most explicit data buyer in this batch: it runs a public Voice Data Partnerships programme that acquires conversations, audiobooks, character performances and news audio for base-model training and fine-tuning, and it staffs Data Ops to translate research needs into labelling specs. It is audio, not visual — but it is the clearest published template for how a creative lab industrialises human data.
What they run now
Core audio models and the research organisation. Below him, Data Ops owns the end-to-end data acquisition and labelling lifecycle: translating research requirements into data specifications and workflows, overseeing in-house and third-party labelling teams, setting guidelines and QC, and managing external data vendors. The Data Operations role is remote-global with offices in New York, London and Warsaw. A Voice Curator role, where staff judged voice naturalness and expressiveness, was removed in Jan 2026.
What they buy
Evidenced at organisational level: licensed voice and audio corpora through Voice Data Partnerships, plus in-house and third-party labelling of audio managed by Data Ops. Inference: perceptual preference judgements on naturalness, expressiveness and performance quality — the Voice Curator role shows they judge this internally — and benchmark/eval design, which research-engineer postings cite. External vendors are referenced in job text but never named.
Career
- 2022 – presentCo-founder and CTO, ElevenLabs