ML Workbench Overview

The ML Workbench (frontend/ml-workbench/) provides 41 components for the complete machine learning lifecycle: AutoML wizards, visual DNN building, experiment tracking, model registry, deployment management, monitoring, feature stores, and Jupyter-style notebooks.

Architecture

ml-workbench/src/
  components/
    AutoML/              # AutoMLWizard, DatasetStep, TargetStep, AlgorithmStep
    DNNBuilder/          # LayerPalette, NetworkGraph, ConversationalBuilder, CodeEditor
    MLflow/              # ExperimentTracker, ModelRegistry (MLflow integration)
    Deployment/          # DeploymentDashboard, EndpointHealthCard, RollbackModal
    Monitoring/          # ModelMonitor
    Training/            # CostWidget, CostBreakdown
    Workspace/           # FeatureStore, NotebookEditor
    advanced/            # ExperimentCompare, MetricsVisualization, ModelVersioning
    AdvancedMonitoring/  # Advanced monitoring dashboard
    ABTesting/           # A/B testing configuration
    DataProfiler/        # Dataset profiling
    DeploymentManager/   # Deployment orchestration
    ExperimentTracker/   # Experiment management
    Explainability/      # Model explainability (SHAP, LIME)
    FeatureStore/        # Feature engineering
    HyperparameterTuning/ # HPO configuration
    MLAssistant/         # AI assistant for ML
    ModelMonitor/        # Real-time model monitoring
    ModelRegistry/       # Model catalog
    Notebooks/           # Notebook editor
    PipelineBuilder/     # ML pipeline construction
  hooks/
    useDNNWebSocket.ts   # WebSocket for DNN builder
  stores/
    dnnBuilderStore.ts   # DNN builder state (Zustand)
  types/
    dnnBuilder.ts        # DNN type definitions

Key Features

Feature	Section	Description
AutoML Wizard	5-step wizard for automated ML	Dataset, target, algorithm, training, review
DNN Builder	Visual neural network design	Layer palette, graph editor, code generation
Model Registry	MLflow-integrated model catalog	Versioning, staging, metadata
Experiment Tracking	MLflow experiment management	Runs, metrics, comparison
Deployment	Model deployment management	Endpoints, health, rollback
Monitoring	Real-time model monitoring	Drift, performance, anomalies
Training	Training lifecycle and costs	GPU costs, training curves
Notebooks	Jupyter-style editor	Cells, execution, versioning

ML Workbench AutoML Wizard