Hi! 馃憢
I'm Vadims Casecnikovs
Distilling human experience into model weights.
What I'm building
- Featured researchSilent SpeechTalking to computers without making a sound, using ultrasound video of the tongue.15.6% WER 路 50 hours of data 路 Generalizes to unseen speakers 路 Published with Aleph Neuro
- Vox Lab12PB of high-quality multimodal training dataUsed by top-5 AI companies and frontier labs 路 Featured in Reuters, The Insider & USENIX Security