CLERC · WHO'S BUILDING CLERC
Florian Méloux
Founder & CEO, CLERC
CLERC builds the data layer for sign language AI, turning two hundred years of spatial grammar into the training data multimodal models are missing.
01 / STORY
I'm Deaf. I grew up between two worlds, the Deaf one and the hearing one, and each gave me a language: sign and speech. I move fluently through both, and that became my edge: I can stand between a Deaf signer and a hearing engineer and bridge not just their words, but their worlds.
From there, I saw what the industry keeps missing. Everyone chases the model. The real bottleneck is the data, and for sign language that data does not exist.
This is not a small gap. Sign language is the richest spatial and temporal language we have: two hundred years of grammar built in three dimensions. As AI turns multimodal and embodied, that is exactly the data it lacks, and no one is building it at the quality models need.
So I am. CLERC is the sign language data infrastructure that universities and AI teams build on top of. We capture, structure, annotate, and license it to the teams training the next generation of models. The model race made this data valuable. I spent a decade getting ready to build it.
02 / THE JOURNEY
Keia: automatic sign language translation.
As Chief of Business, I helped deploy a French Sign Language avatar across the French government, Airbus, Thales, and TotalEnergies. I saw how much the world wants this, and how fast it breaks without the right data.
Spokhand: a nonprofit for data & visibility.
We brought sign language to the public and reached 15M+ views in 18 months, then partnered with the Tokyo 2025 Deaflympics to put Deaf athletes, and the technology, in front of the world. It proved the demand. It also exposed the bottleneck: the data was not there.
03 / EDUCATION
HEC Paris · UC Berkeley
04 / PRESS & RESEARCH
ARTICLES
Building the data layer for sign language AI.
Researcher, investor, or partner? Let's talk.