Skyulf is a local-first MLOps workspace. Visual canvas for feature engineering, model training with hyperparameter search, experiment tracking, and one-click testing.

It functions like a visual automation tool (like n8n) but for ML pipelines. You drag-and-drop nodes to handle data loading, preprocessing (25+ nodes), feature engineering, and model training. No code needed for common tasks.

What's next? I am working on integrating powerful models like XGBoost/LightGBM/CatBoost, adding SHAP/LIME explainability, and eventually building a visual LLM builder (LangChain nodes) and more EDA features.

It's in active alpha. It works, but expect bugs or incomplete features.

-- I'd love feedback. Does the visual MLOps tool solve a problem for you? What’s the first custom node or feature you'd look for? or What you would need in this app to be more useful for you.

Thanks for checking it out!