Why We're Funding Maziyar Panahi
Maziyar Panahi is building the open foundation for medical AI. As founder of OpenMed and a Hugging Face PRO contributor with 16+ years between France's CNRS and enterprise NLP, he's making sure hospitals, researchers, and regulated industries have auditable, sovereign models they can deploy on their own infrastructure—not locked behind proprietary APIs.
#1
Most Referenced Org
OpenMed on Hugging Face, Spring 2026
3,800+
Models Published
Combined across his HF profile and OpenMed org
150M+
Spark NLP Downloads
Ecosystem he helped scale across enterprise
Major Open Source Contributions
An open-source medical AI initiative spanning biomedical NER, clinical reasoning, and healthcare-specific LLMs. Built to be auditable and deployable inside hospitals—HIPAA and GDPR-aware, with no vendor lock-in. Recognized by Hugging Face as the most referenced organization powering open-source AI research.
Clinical NER
Biomedical LLMs
HIPAA-aware
GDPR-aware
On-Prem Deploy
A family of post-trained models including Calme-3 in 78B and 3B sizes, advancing open-weight French-language capabilities. Built with modern post-training methods—SFT, preference modeling, and GRPO—and published openly for fine-tuning and downstream research.
SFT
Preference Modeling
GRPO
Model Merging
End-to-end protein AI pipeline covering structure prediction, sequence design, and codon optimization. Four production models trained in 55 GPU-hours, with CodonRoBERTa-large-v2 hitting a perplexity of 4.10—demonstrating that frontier biological AI doesn't require frontier-lab budgets.
One of Hugging Face's most prolific publishers of GGUF, GPTQ, and AWQ quantizations—making the latest open-weight models runnable on consumer hardware within hours of release. The plumbing that lets builders everywhere actually use what gets released.
Technical Excellence
National-Scale Infrastructure: 14 years at France's CNRS / ISC-PIF architecting AI platforms—360B+ records, 140+ servers, multi-cloud deployments serving public research.
State-of-the-Art Medical NER: Built clinical transformer models with +9.7pp improvements over prior benchmarks on biomedical entity recognition tasks.
Enterprise Scale: Led the Spark NLP ecosystem to 150M+ downloads and built GenAI solutions for healthcare and pharmaceutical clients operating under strict regulatory regimes.
