Run models. Edit contours. Ship faster.
MLUI is the universal ML inference platform where researchers run models on images, annotators refine contour outputs, and admins build custom workspace layouts -- all without writing frontend code.
Input
input node
Model
model node
Viewer
viewer node
Platform capabilities
Everything you need to ship ML faster
From custom workspace layouts to interactive annotation editors -- MLUI gives your team a complete visual toolchain for ML inference.
Customizable Canvas
Drag, resize, and snap functional nodes onto a dot-grid canvas. Connect inputs, models, and viewers to build your exact inference workflow -- no code required.
REQ-4Contour Editor
Overlay ML-generated contours on medical images and refine them directly on the canvas. Drag vertices, insert points, toggle layers, and save with optimistic locking.
REQ-5Dynamic Button Panel
Import a JSON config or add buttons manually to generate action triggers for any model. Highlight specific contour layers and invoke backend libraries in one click.
REQ-6NextAuth.js
Google OAuth & Magic Link
Role-Based Access
Admin & User roles
File Manager
Upload, browse, delete
Inference Engine
Async subprocess queue
Pricing
Simple, transparent pricing
Start free and scale as your inference workload grows.
Free
Perfect for solo researchers and quick prototypes.
- Basic visualization & contour editor
- 1 active project
- Local model execution
- Community support
- Core workspace canvas
Pro
Built for teams running models in production.
- Full canvas layout editor
- Unlimited active projects
- Hosted Python backend
- Up to 5 team members
- Priority support
- Dynamic button importer
- Advanced contour tools
Enterprise
Dedicated resources for large-scale annotation teams.
- Custom model integrations
- Dedicated compute resources
- SSO & advanced security
- Unlimited team members
- SLA guarantee
- Custom onboarding
- Audit logs & compliance
FAQ
Common questions
What is MLUI?+
MLUI is a universal web-based machine learning inference platform that allows users to run models on images, edit contour JSON outputs on an interactive canvas, and build custom widget layouts -- all without writing custom frontend code.
How does the layout engine work?+
MLUI provides a free-form dot-grid canvas. Admins can drag, drop, and resize widget nodes like the File Manager, Image Preview, and Button Panel. Nodes are connected by drawing edges between output and input ports, creating a visual data flow pipeline.
Can I run my own custom Python scripts?+
Yes! Admins can upload custom Python model files, manage virtual environments via the built-in venv builder, and trigger inference directly from the admin panel. The system streams real-time build and execution logs back to the browser.
Is my data secure?+
Absolutely. All uploaded images and inference results are stored in your dedicated MongoDB database with strict access control. Access is controlled through role-based authentication using NextAuth.js with Google OAuth and Magic Link support.
Ready to run your first model?
Join hundreds of ML researchers and annotators already using MLUI to visualize and refine model outputs.