Built by clinicians, for clinicians
We started Meditopia because clinical AI models were disappearing into PDF appendices, locked corporate servers, and one-off GitHub repos nobody could find or trust. Healthcare deserves a platform that takes compliance, reproducibility, and open science as seriously as the research itself.
4,200+
Models published
890+
Datasets shared
12k+
Researchers
60+
Institutions
Open by default
Every public model and dataset is freely accessible to any researcher, anywhere. No paywalls, no access request queues for open-science work.
Compliant by design
HIPAA, GDPR, FDA 510(k) guidance, and FHIR R4 aren't checklists we added later. They are the product foundation.
Community first
We are building the infrastructure layer for clinical AI together with the research community — not for it.
Our enterprise services
Every collaboration scenario between health organisations has a different trust model. Meditopia is purpose-built to handle all of them securely.
Secure Federated Fine-Tuning
Health Org 1 shares encrypted patient data, usage instructions, and sample data with Meditopia's secure enclave. Health Org 2 submits a base model. Fine-tuning runs entirely inside the enclave — raw data is never exposed. Only trained weights leave. All temporary data is purged.
The team
Clinicians, engineers, and researchers who got tired of the status quo.
Amir Mirmehrkar
Co-founder & CEO
Former ML researcher at Stanford Medicine. Frustrated that clinical AI models lived in paper appendices — so she built a home for them.
Pedram Hashemi
Co-founder & CTO
Built ML infrastructure at scale. Believes the biggest bottleneck in clinical AI is not the models — it's reproducibility and access.
Our clients
From flagship academic medical centres to independent research labs — these are the institutions that have made Meditopia part of their clinical AI workflow.
Mayo Clinic
Academic Medical Centre
Using Meditopia for cardiology and oncology model distribution across 20 research sites.
Karolinska University Hospital
University Hospital
Nordic flagship for compliant clinical AI sharing with GDPR-native dataset hosting.
Gates Medical Research Institute
Research Institute
Publishing global health AI models and EHR datasets to the open-source community.
IIT Bombay — Biomedical Informatics
University Lab
Student researchers publishing reproducible clinical AI for their dissertations.
Charité Berlin
University Hospital
Germany's largest university hospital using Meditopia for regulatory-ready model cards.
University of Tokyo Hospital
Academic Medical Centre
Radiology AI lab publishing benchmark-tracked models with full reproducibility packages.
Cited by the community
Researchers across academia, industry, and clinical practice share their experience building on Meditopia.
Meditopia has become the standard repository for our radiology AI lab. We upload every model release there, and the download metrics give us real signal on community adoption.
Prof. Yuki Tanaka
Director, Medical AI Lab · University of Tokyo Hospital
The compliance tooling is what sold us. Having HIPAA badges and DUA gating built into the platform removed months of internal legal review before we could share our EHR models.
Dr. Amara Diallo
Senior Research Scientist · Gates Medical Research Institute
We migrated our entire model registry to Meditopia. The CLI is smooth, versioning works exactly like GitHub, and the benchmark tracking saved us a custom internal dashboard.
Erik Lindström
ML Platform Lead · Karolinska University Hospital
For a small academic lab with no DevOps budget, having a production-grade model hub for free is remarkable. Our students publish reproducible clinical AI research in minutes.
Dr. Priya Nair
Assistant Professor, Biomedical Informatics · IIT Bombay
The dataset discovery features saved our team weeks — we found three compliant chest X-ray datasets we didn't know existed, all with DOIs and license metadata intact.
Dr. Carlos Medina
Chief Data Officer · Mayo Clinic Platform
Meditopia is the missing piece between "we trained a model" and "the community can actually use it." The inference API and FHIR adapter saved us 3 months of engineering.
Dr. Lena Fischer
Head of AI · Charité — Universitätsmedizin Berlin
Join the community
Publish your clinical AI models and datasets where the world can find them.