Add Skyulf to graph-based ML tools list

[flyingriverhorse]

What is this tool for?

Skyulf is an open-source MLOps platform and a standalone Python library (skyulf-core) for building leakage-proof, stateless ML pipelines. It separates training logic from inference, allowing models to be saved as version-agnostic JSON artifacts instead of problematic Python pickles. It also includes a Polars-powered automated EDA engine.

Differences?

• vs Scikit-learn: Skyulf pipelines are stateless (JSON deployment) and architecturally enforce leakage prevention (SplitDataset), whereas Scikit-learn relies on user discipline and heavy pickles.
• vs MLflow: Skyulf builds the pipeline logic and architecture; MLflow tracks the results.
• vs Cloud: Skyulf is local-first/self-hosted, requiring no data upload.