Mahidol University · High Performance Computing

MUSCULAR
Cluster

Mahidol University Special Computing Unit for Life-science & Advanced Research. Platform supporting life-science, medical analytics, AI research, and large-scale computation. Designed for secure private workspaces, federated learning collaboration, and international research networks.

Apply to Use Explore Platform

01 — About

What is MUSCULAR?

MUSCULAR is the second phase of Mahidol University's HPC initiative, expanding GPU infrastructure from NVIDIA A100 to H200 and introducing a unified management platform for secure, isolated research workspaces. It supports clinical AI, genomics, large-scale simulation, and federated learning across Mahidol's network — without moving sensitive data.

For a full overview of the platform infrastructure — compute nodes, storage fabric, local AI models, and cloud integrations — see the MUSCULAR system architecture →

Who We Serve

Researchers & Clinical Teams

Principal investigators, data scientists, and clinical teams across Mahidol University and its affiliated hospitals. Open to international research partners through formal collaboration agreements.

Life Sciences Medical AI Genomics Imaging Clinical Data
Governance

Request-Based & Auditable

All access is project-based and requires approval. Each project receives an isolated container workspace — no data mixing between users. Allocations specify GPU, RAM, storage, and time window for full auditability and regulatory compliance.

Isolated Workspaces Auditable Compliant

02 — Infrastructure

What We Provide

Infrastructure and services available to approved research projects.

Platform Service

Federated Learning Engine

Train AI models jointly across organisations without moving sensitive data. Only model parameters are exchanged — data stays at its source.

Software Stack

Pre-configured Environments

Container images with common ML frameworks, bioinformatics tools, and research libraries — ready on first login. Additional frameworks available on request.

Collaboration

International Research Network

Formal pathways for cross-border federated research. Data stays in country; only model gradients travel across borders. Contact us to discuss international collaboration arrangements.

Secure Access

Secure Web Access

Access your workspace directly from any web browser — no VPN installation, no command-line setup required. Built on browser isolation and Zero Trust principles, your session runs in a secure remote environment. Nothing is executed on your local device.

Support

Technical Assistance

Onboarding support, Slurm scheduling guidance, and a dedicated HPC advisor assigned to each project.

  • Slurm job scheduler
  • HPC advisor per project
  • Documentation portal access

03 — Request Access

Apply to Use MUSCULAR

All usage is project-based and requires approval. Download the application form, complete it, and submit by email to the MUSCULAR team.

How the process works

DDownload and complete the PDF application form. Attach any supporting documents that help us understand your project — such as a research proposal, ethics approval, prior publications, or preliminary results. Email your completed form and attachments to the MUSCULAR team. The allocation committee reviews all applications within 5–7 business days.

01Download the PDF application form below
02Complete all sections and obtain PI signature
03Email the completed form to consult@mahidol.ac.th
04Review by MUSCULAR allocation committee (5–7 days)
05Workspace provisioned & onboarding call scheduled
Submit completed forms to:
consult@mahidol.ac.th
MUSCULAR HPC Application Form
PDF · A4 · English
Applicant & faculty details
Project title & abstract
Research domain & ethics status
GPU nodes, RAM & storage request
Requested duration & start date
Frameworks & software needed
Data description & sensitivity
Collaborating institutions
Principal Investigator signature
Download Application Form (PDF)

For questions about the application process or eligibility, contact consult@mahidol.ac.th before submitting.