Documentation
Fendlemark documentation
Everything you need to run Fendlemark in front of your ML platform, from the first go install to writing your own adapter. Documentation for v0.9.5.
Start here
Getting started
Install the CLI, connect MLflow and Argo, open the console. About ten minutes.
Architecture
The control plane, the four adapters, the lattice, and the console, and how a request moves between them.
Configuration
Every key in fendlemark.yaml with its type and default.
Integrations
MLflow
Experiments, runs, metrics, artifacts, and the Model Registry over the tracking server API.
Metaflow
Flows, runs, steps, and tasks from the metadata service.
Argo Workflows
Workflows, templates, and cron workflows from Argo Server or a direct Kubernetes watch.
Kubeflow
Pipelines, recurring runs, Katib experiments, and Notebook servers.
Reference
Writing an adapter
The six-method interface, facets, correlation keys, and the conformance suite.
The console
Views, facet panels, the @fendlemark/ui component library, and theming.
Authentication
Fendlemark Respiratory Identity: enrolment, multi-factor, troubleshooting, error codes.
CLI reference
init, serve, adapters, lattice, auth, doctor, version.
Go API reference
Packages, interfaces, the HTTP API, and the stability policy.
Migrating 0.8 to 0.9
One renamed key, documented in proportion to the trust it asks of you.
Packages
The Go module is fendlemark.app/fendlemark and the Vue 3 component library is @fendlemark/ui. Install commands, package layout, and links to the source are on the Packages page.