Docs

Documentation for both LydianAI open-source projects.

Federated ML

Set up the coordinator and workers, configure GPU paths, and run your first training round.

Server/coordinator + worker loop for FedAvg training over Tailscale.

Run the FedAvg training PoC: start server, start workers, launch a training run, and monitor results.

Put server and workers on the same tailnet and use stable 100.x IPs.

Modern CUDA wheels vs Pascal (GTX 1080/1080 Ti) pinned legacy stack.

Start runs, monitor progress, and fetch results using client.submit_job.

Deploy the backend, create a SW System project, and manage compliance artifacts for UNECE R155, R156, and ISO 29119.

Deploy the Assurance-as-Code backend using Docker Compose.

Set up a SW System project with the applicable ASIL level, Automotive SPICE capability target, and standards profile.

Create requirements, hazard analyses, test cases, and test results. Link them to form a complete traceability chain.

Create a baseline, run the release-gate check, approve, and release — with severity-based blocking and stale-gate protection.

Set up the GitHub App, configure a repository connector, and add .aac/repo.yml to link source code to requirements automatically.

The four roles in Assurance-as-Code and the per-operation authorization matrix.

How Assurance-as-Code flags downstream artefacts when a released item changes, and how to clear a suspect state.

The aac command-line client — full parity with the web console for import, validation, traceability, baseline, and release-gate workflows.