v0.1.0 - Initial public release

MLIS v0.1.0 is the first public release of the Machine Learning Infrastructure Service.

MLIS is a local-first AI infrastructure reference implementation for durable inference jobs. It is built to make the control-plane side of AI systems concrete and runnable, including:

• job submission and durable state
• scheduler and worker separation
• lease-based recovery
• tenant-scoped authorization
• artifact-backed inputs and outputs
• React operator console and CLI workflows
• observability and audit signals

What this release demonstrates:

• a local Docker Compose demo path
• one API/control-plane service plus multiple worker/data-plane services
• recovery when a worker dies mid-run
• simulated GPU-aware scheduling without requiring a physical GPU

Recommended first demo:

1. Start the stack with Docker Compose.
2. Open /console.
3. Submit a sleep or gpu_demo job.
4. Kill the active worker and watch the job get reassigned.
5. Confirm the job completes successfully.

This release also includes a lease recovery demo video showing a long-running job being recovered after its active worker is terminated.